Add buffer donation to benchmark

.buildkite/check-wheel-size.py

@@ -1,43 +0,0 @@
-import os
-import sys
-import zipfile
-
-# Read the VLLM_MAX_SIZE_MB environment variable, defaulting to 250 MB
-VLLM_MAX_SIZE_MB = int(os.environ.get('VLLM_MAX_SIZE_MB', 250))
-
-
-def print_top_10_largest_files(zip_file):
-    """Print the top 10 largest files in the given zip file."""
-    with zipfile.ZipFile(zip_file, 'r') as z:
-        file_sizes = [(f, z.getinfo(f).file_size) for f in z.namelist()]
-        file_sizes.sort(key=lambda x: x[1], reverse=True)
-        for f, size in file_sizes[:10]:
-            print(f"{f}: {size / (1024 * 1024):.2f} MBs uncompressed.")
-
-
-def check_wheel_size(directory):
-    """Check the size of .whl files in the given directory."""
-    for root, _, files in os.walk(directory):
-        for file_name in files:
-            if file_name.endswith(".whl"):
-                wheel_path = os.path.join(root, file_name)
-                wheel_size_mb = os.path.getsize(wheel_path) / (1024 * 1024)
-                if wheel_size_mb > VLLM_MAX_SIZE_MB:
-                    print(f"Not allowed: Wheel {wheel_path} is larger "
-                          f"({wheel_size_mb:.2f} MB) than the limit "
-                          f"({VLLM_MAX_SIZE_MB} MB).")
-                    print_top_10_largest_files(wheel_path)
-                    return 1
-                else:
-                    print(f"Wheel {wheel_path} is within the allowed size "
-                          f"({wheel_size_mb:.2f} MB).")
-    return 0
-
-
-if __name__ == "__main__":
-    if len(sys.argv) < 2:
-        print("Usage: python check-wheel-size.py <directory>")
-        sys.exit(1)
-
-    directory = sys.argv[1]
-    sys.exit(check_wheel_size(directory))

.buildkite/download-images.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+set -ex
+set -o pipefail
+
+(which wget && which curl) || (apt-get update && apt-get install -y wget curl)
+
+# aws s3 sync s3://air-example-data-2/vllm_opensource_llava/ images/
+mkdir -p images
+cd images
+wget https://air-example-data-2.s3.us-west-2.amazonaws.com/vllm_opensource_llava/stop_sign_pixel_values.pt
+wget https://air-example-data-2.s3.us-west-2.amazonaws.com/vllm_opensource_llava/stop_sign_image_features.pt
+wget https://air-example-data-2.s3.us-west-2.amazonaws.com/vllm_opensource_llava/cherry_blossom_pixel_values.pt
+wget https://air-example-data-2.s3.us-west-2.amazonaws.com/vllm_opensource_llava/cherry_blossom_image_features.pt
+wget https://air-example-data-2.s3.us-west-2.amazonaws.com/vllm_opensource_llava/stop_sign.jpg
+wget https://air-example-data-2.s3.us-west-2.amazonaws.com/vllm_opensource_llava/cherry_blossom.jpg
+
+cd -

.buildkite/lm-eval-harness/configs/DeepSeek-V2-Lite-Chat.yaml

@@ -1,12 +0,0 @@
-# bash ./run-lm-eval-gsm-vllm-baseline.sh -m deepseek-ai/DeepSeek-V2-Lite-Chat -b "auto" -l 1000 -f 5 -t 2
-model_name: "deepseek-ai/DeepSeek-V2-Lite-Chat"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.671
-  - name: "exact_match,flexible-extract"
-    value: 0.664
-limit: 1000
-num_fewshot: 5
-trust_remote_code: True

.buildkite/lm-eval-harness/configs/Meta-Llama-3-70B-Instruct-FBGEMM-nonuniform.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh -m nm-testing/Meta-Llama-3-70B-Instruct-FBGEMM-nonuniform -b auto -l 1000 -f 5
-model_name: "nm-testing/Meta-Llama-3-70B-Instruct-FBGEMM-nonuniform"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.905
-  - name: "exact_match,flexible-extract"
-    value: 0.905
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-70B-Instruct.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh -m meta-llama/Meta-Llama-3-70B-Instruct -b 32 -l 250 -f 5
-model_name: "meta-llama/Meta-Llama-3-70B-Instruct"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.892
-  - name: "exact_match,flexible-extract"
-    value: 0.892
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct-Channelwise-compressed-tensors.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Meta-Llama-3-8B-Instruct-W8A8-FP8-Channelwise-compressed-tensors -b auto -l 1000 -f 5 -t 1
-model_name: "nm-testing/Meta-Llama-3-8B-Instruct-W8A8-FP8-Channelwise-compressed-tensors"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.752
-  - name: "exact_match,flexible-extract"
-    value: 0.754
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct-FBGEMM-nonuniform.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Meta-Llama-3-8B-Instruct-FBGEMM-nonuniform -b auto -l 1000 -f 5 -t 1
-model_name: "nm-testing/Meta-Llama-3-8B-Instruct-FBGEMM-nonuniform"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.753
-  - name: "exact_match,flexible-extract"
-    value: 0.753
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct-FP8-compressed-tensors.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Meta-Llama-3-8B-FP8-compressed-tensors-test -b 32 -l 1000 -f 5 -t 1
-model_name: "nm-testing/Meta-Llama-3-8B-FP8-compressed-tensors-test"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.755
-  - name: "exact_match,flexible-extract"
-    value: 0.755
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct-FP8.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m neuralmagic/Meta-Llama-3-8B-Instruct-FP8 -b 32 -l 250 -f 5 -t 1
-model_name: "neuralmagic/Meta-Llama-3-8B-Instruct-FP8"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.753
-  - name: "exact_match,flexible-extract"
-    value: 0.753
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct-INT8-compressed-tensors.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Meta-Llama-3-8B-Instruct-W8-Channel-A8-Dynamic-Per-Token-Test -b "auto" -l 250 -f 5 -t 1
-model_name: "nm-testing/Meta-Llama-3-8B-Instruct-W8-Channel-A8-Dynamic-Per-Token-Test"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.728
-  - name: "exact_match,flexible-extract"
-    value: 0.728
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct-nonuniform-compressed-tensors.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Meta-Llama-3-8B-Instruct-nonuniform-test -b auto -l 1000 -f 5 -t 1
-model_name: "nm-testing/Meta-Llama-3-8B-Instruct-nonuniform-test"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.758
-  - name: "exact_match,flexible-extract"
-    value: 0.759
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh -m meta-llama/Meta-Llama-3-8B-Instruct -b 32 -l 250 -f 5 -t 1
-model_name: "meta-llama/Meta-Llama-3-8B-Instruct"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.756
-  - name: "exact_match,flexible-extract"
-    value: 0.752
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-QQQ.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m HandH1998/QQQ-Llama-3-8b-g128 -b 32 -l 1000 -f 5 -t 1
-model_name: "HandH1998/QQQ-Llama-3-8b-g128"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.419
-  - name: "exact_match,flexible-extract"
-    value: 0.416
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Minitron-4B-Base-FP8.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m mgoin/Minitron-4B-Base-FP8 -b auto -l 1000 -f 5 -t 1
-model_name: "mgoin/Minitron-4B-Base-FP8"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.233
-  - name: "exact_match,flexible-extract"
-    value: 0.236
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Mixtral-8x22B-Instruct-v0.1-FP8-Dynamic.yaml

@@ -1,11 +0,0 @@
-# bash ./run-lm-eval-gsm-vllm-baseline.sh -m neuralmagic/Mixtral-8x22B-Instruct-v0.1-FP8-dynamic -b "auto" -l 250 -f 5 -t 8
-model_name: "neuralmagic/Mixtral-8x22B-Instruct-v0.1-FP8-dynamic"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.86
-  - name: "exact_match,flexible-extract"
-    value: 0.86
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Mixtral-8x7B-Instruct-v0.1-FP8.yaml

@@ -1,11 +0,0 @@
-# bash ./run-lm-eval-gsm-vllm-baseline.sh -m neuralmagic/Mixtral-8x7B-Instruct-v0.1-FP8 -b "auto" -l 250 -f 5 -t 4
-model_name: "neuralmagic/Mixtral-8x7B-Instruct-v0.1-FP8"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.624
-  - name: "exact_match,flexible-extract"
-    value: 0.624
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Mixtral-8x7B-Instruct-v0.1.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh -m neuralmagic/Mixtral-8x7B-Instruct-v0.1 -b 32 -l 250 -f 5 -t 4
-model_name: "mistralai/Mixtral-8x7B-Instruct-v0.1"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.616
-  - name: "exact_match,flexible-extract"
-    value: 0.632
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Qwen2-1.5B-Instruct-FP8W8.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Qwen2-1.5B-Instruct-FP8W8 -b auto -l 1000 -f 5 -t 1
-model_name: "nm-testing/Qwen2-1.5B-Instruct-FP8W8"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.578
-  - name: "exact_match,flexible-extract"
-    value: 0.585
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Qwen2-1.5B-Instruct-INT8-compressed-tensors.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m neuralmagic/Qwen2-1.5B-Instruct-quantized.w8a8 -b "auto" -l 1000 -f 5 -t 1
-model_name: "neuralmagic/Qwen2-1.5B-Instruct-quantized.w8a8"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.593
-  - name: "exact_match,flexible-extract"
-    value: 0.588
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Qwen2-1.5B-Instruct-W8A16-compressed-tensors.yaml

@@ -1,11 +0,0 @@
-# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m nm-testing/Qwen2-1.5B-Instruct-W8A16-Channelwise -b "auto" -l 1000 -f 5 -t 1
-model_name: "nm-testing/Qwen2-1.5B-Instruct-W8A16-Channelwise"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.595
-  - name: "exact_match,flexible-extract"
-    value: 0.582
-limit: 1000
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/Qwen2-57B-A14-Instruct.yaml

@@ -1,11 +0,0 @@
-# bash ./run-lm-eval-gsm-vllm-baseline.sh -m Qwen/Qwen2-57B-A14B-Instruct -b "auto" -l 250 -f 5 -t 4
-model_name: "Qwen/Qwen2-57B-A14B-Instruct"
-tasks:
-- name: "gsm8k"
-  metrics:
-  - name: "exact_match,strict-match"
-    value: 0.792
-  - name: "exact_match,flexible-extract"
-    value: 0.824
-limit: 250
-num_fewshot: 5

.buildkite/lm-eval-harness/configs/models-large.txt

@@ -1,5 +0,0 @@
-Meta-Llama-3-70B-Instruct-FBGEMM-nonuniform.yaml
-Meta-Llama-3-70B-Instruct.yaml
-Mixtral-8x7B-Instruct-v0.1.yaml
-Qwen2-57B-A14-Instruct.yaml
-DeepSeek-V2-Lite-Chat.yaml

.buildkite/lm-eval-harness/configs/models-small.txt

@@ -1,9 +0,0 @@
-Meta-Llama-3-8B-Instruct.yaml
-Meta-Llama-3-8B-Instruct-FP8-compressed-tensors.yaml
-Meta-Llama-3-8B-Instruct-INT8-compressed-tensors.yaml
-Meta-Llama-3-8B-Instruct-nonuniform-compressed-tensors.yaml
-Meta-Llama-3-8B-Instruct-Channelwise-compressed-tensors.yaml
-Minitron-4B-Base-FP8.yaml
-Qwen2-1.5B-Instruct-INT8-compressed-tensors.yaml
-Qwen2-1.5B-Instruct-FP8W8.yaml
-Meta-Llama-3-8B-QQQ.yaml

.buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh

@@ -1,46 +0,0 @@
-#!/bin/bash
-# We can use this script to compute baseline accuracy on GSM for transformers.
-#
-# Make sure you have lm-eval-harness installed:
-#   pip install git+https://github.com/EleutherAI/lm-evaluation-harness.git@9516087b81a61d0e220b22cc1b75be76de23bc10
-
-usage() {
-    echo``
-    echo "Runs lm eval harness on GSM8k using huggingface transformers."
-    echo "This pathway is intended to be used to create baselines for "
-    echo "our automated nm-test-accuracy workflow"
-    echo
-    echo "usage: ${0} <options>"
-    echo
-    echo "  -m    - huggingface stub or local directory of the model"
-    echo "  -b    - batch size to run the evaluation at"
-    echo "  -l    - limit number of samples to run"
-    echo "  -f    - number of fewshot samples to use"
-    echo
-}
-
-while getopts "m:b:l:f:" OPT; do
-  case ${OPT} in
-    m ) 
-        MODEL="$OPTARG"
-        ;;
-    b ) 
-        BATCH_SIZE="$OPTARG"
-        ;;
-    l ) 
-        LIMIT="$OPTARG"
-        ;;
-    f ) 
-        FEWSHOT="$OPTARG"
-        ;;
-    \? ) 
-        usage
-        exit 1
-        ;;
-  esac
-done
-
-lm_eval --model hf \
-  --model_args pretrained=$MODEL,parallelize=True \
-  --tasks gsm8k --num_fewshot $FEWSHOT --limit $LIMIT \
-  --batch_size $BATCH_SIZE

.buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh

@@ -1,51 +0,0 @@
-#!/bin/bash
-# We can use this script to compute baseline accuracy on GSM for vllm.
-# We use this for fp8, which HF does not support.
-#
-# Make sure you have lm-eval-harness installed:
-#   pip install lm-eval==0.4.3
-
-usage() {
-    echo``
-    echo "Runs lm eval harness on GSM8k using huggingface transformers."
-    echo "This pathway is intended to be used to create baselines for "
-    echo "our automated nm-test-accuracy workflow"
-    echo
-    echo "usage: ${0} <options>"
-    echo
-    echo "  -m    - huggingface stub or local directory of the model"
-    echo "  -b    - batch size to run the evaluation at"
-    echo "  -l    - limit number of samples to run"
-    echo "  -f    - number of fewshot samples to use"
-    echo "  -t    - tensor parallel size to run at"
-    echo
-}
-
-while getopts "m:b:l:f:t:" OPT; do
-  case ${OPT} in
-    m ) 
-        MODEL="$OPTARG"
-        ;;
-    b ) 
-        BATCH_SIZE="$OPTARG"
-        ;;
-    l ) 
-        LIMIT="$OPTARG"
-        ;;
-    f ) 
-        FEWSHOT="$OPTARG"
-        ;;
-    t )
-        TP_SIZE="$OPTARG"
-        ;;
-    \? ) 
-        usage
-        exit 1
-        ;;
-  esac
-done
-
-lm_eval --model vllm \
-  --model_args pretrained=$MODEL,tensor_parallel_size=$TP_SIZE,distributed_executor_backend="ray",trust_remote_code=true,max_model_len=4096 \
-  --tasks gsm8k --num_fewshot $FEWSHOT --limit $LIMIT \
-  --batch_size $BATCH_SIZE

.buildkite/lm-eval-harness/run-tests.sh

@@ -1,59 +0,0 @@
-#!/bin/bash
-
-usage() {
-    echo``
-    echo "Runs lm eval harness on GSM8k using vllm and compares to "
-    echo "precomputed baseline (measured by HF transformers.)"
-    echo
-    echo "usage: ${0} <options>"
-    echo
-    echo "  -c    - path to the test data config (e.g. configs/small-models.txt)"
-    echo "  -t    - tensor parallel size"
-    echo
-}
-
-SUCCESS=0
-
-while getopts "c:t:" OPT; do
-  case ${OPT} in
-    c ) 
-        CONFIG="$OPTARG"
-        ;;
-    t )
-        TP_SIZE="$OPTARG"
-        ;;
-    \? )
-        usage
-        exit 1
-        ;;
-  esac
-done
-
-# Parse list of configs.
-IFS=$'\n' read -d '' -r -a MODEL_CONFIGS < $CONFIG
-
-for MODEL_CONFIG in "${MODEL_CONFIGS[@]}"
-do
-    LOCAL_SUCCESS=0
-    
-    echo "=== RUNNING MODEL: $MODEL_CONFIG WITH TP SIZE: $TP_SIZE==="
-
-    export LM_EVAL_TEST_DATA_FILE=$PWD/configs/${MODEL_CONFIG}
-    export LM_EVAL_TP_SIZE=$TP_SIZE
-    pytest -s test_lm_eval_correctness.py || LOCAL_SUCCESS=$?
-
-    if [[ $LOCAL_SUCCESS == 0 ]]; then
-        echo "=== PASSED MODEL: ${MODEL_CONFIG} ==="
-    else
-        echo "=== FAILED MODEL: ${MODEL_CONFIG} ==="
-    fi
-
-    SUCCESS=$((SUCCESS + LOCAL_SUCCESS))
-
-done
-
-if [ "${SUCCESS}" -eq "0" ]; then
-    exit 0
-else
-    exit 1
-fi

.buildkite/lm-eval-harness/test_lm_eval_correctness.py

@@ -1,58 +0,0 @@
-"""
-LM eval harness on model to compare vs HF baseline computed offline.
-Configs are found in configs/$MODEL.yaml
-
-* export LM_EVAL_TEST_DATA_FILE=configs/Meta-Llama-3-70B-Instruct.yaml
-* export LM_EVAL_TP_SIZE=4 
-* pytest -s test_lm_eval_correctness.py
-"""
-
-import os
-from pathlib import Path
-
-import lm_eval
-import numpy
-import yaml
-
-RTOL = 0.05
-TEST_DATA_FILE = os.environ.get(
-    "LM_EVAL_TEST_DATA_FILE",
-    ".buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-Instruct.yaml")
-
-TP_SIZE = os.environ.get("LM_EVAL_TP_SIZE", 1)
-
-
-def launch_lm_eval(eval_config):
-    trust_remote_code = eval_config.get('trust_remote_code', False)
-
-    model_args = f"pretrained={eval_config['model_name']}," \
-                 f"tensor_parallel_size={TP_SIZE}," \
-                 f"add_bos_token=true," \
-                 f"trust_remote_code={trust_remote_code}"
-
-    results = lm_eval.simple_evaluate(
-        model="vllm",
-        model_args=model_args,
-        tasks=[task["name"] for task in eval_config["tasks"]],
-        num_fewshot=eval_config["num_fewshot"],
-        limit=eval_config["limit"],
-        batch_size="auto")
-
-    return results
-
-
-def test_lm_eval_correctness():
-    eval_config = yaml.safe_load(
-        Path(TEST_DATA_FILE).read_text(encoding="utf-8"))
-
-    # Launch eval requests.
-    results = launch_lm_eval(eval_config)
-
-    # Confirm scores match ground truth.
-    for task in eval_config["tasks"]:
-        for metric in task["metrics"]:
-            ground_truth = metric["value"]
-            measured_value = results["results"][task["name"]][metric["name"]]
-            print(f'{task["name"]} | {metric["name"]}: '
-                  f'ground_truth={ground_truth} | measured={measured_value}')
-            assert numpy.isclose(ground_truth, measured_value, rtol=RTOL)

.buildkite/nightly-benchmarks/README.md

@@ -1,153 +0,0 @@
-# vLLM benchmark suite
-
-
-## Introduction
-
-This directory contains two sets of benchmark for vllm.
-- Performance benchmark: benchmark vllm's performance under various workload, for **developers** to gain clarity on whether their PR improves/degrades vllm's performance
-- Nightly benchmark: compare vllm's performance against alternatives (tgi, trt-llm and lmdeploy), for **the public** to know when to choose vllm.
-
-
-See  [vLLM performance dashboard](https://perf.vllm.ai) for the latest performance benchmark results and [vLLM GitHub README](https://github.com/vllm-project/vllm/blob/main/README.md) for latest nightly benchmark results.
-
-
-## Performance benchmark quick overview
-
-**Benchmarking Coverage**: latency, throughput and fix-qps serving on A100 (the support for FP8 benchmark on H100 is coming!), with different models.
-
-**Benchmarking Duration**: about 1hr.
-
-**For benchmarking developers**: please try your best to constraint the duration of benchmarking to about 1 hr so that it won't take forever to run.
-
-
-## Nightly benchmark quick overview
-
-**Benchmarking Coverage**: Fix-qps serving on A100 (the support for FP8 benchmark on H100 is coming!) on Llama-3 8B, 70B and Mixtral 8x7B. 
-
-**Benchmarking engines**: vllm, TGI, trt-llm and lmdeploy.
-
-**Benchmarking Duration**: about 3.5hrs.
-
-
-
-## Trigger the benchmark
-
-Performance benchmark will be triggered when:
-- A PR being merged into vllm.
-- Every commit for those PRs with `perf-benchmarks` label AND `ready` label.
-
-Nightly benchmark will be triggered when:
-- Every commit for those PRs with `perf-benchmarks` label and `nightly-benchmarks` label.
-
-
-
-
-## Performance benchmark details
-
-
-See [performance-benchmarks-descriptions.md](performance-benchmarks-descriptions.md) for detailed descriptions, and use `tests/latency-tests.json`, `tests/throughput-tests.json`, `tests/serving-tests.json` to configure the test cases.
-
-
-#### Latency test
-
-Here is an example of one test inside `latency-tests.json`:
-
-```json
-[
-    {
-        "test_name": "latency_llama8B_tp1",
-        "parameters": {
-            "model": "meta-llama/Meta-Llama-3-8B",
-            "tensor_parallel_size": 1,
-            "load_format": "dummy",
-            "num_iters_warmup": 5,
-            "num_iters": 15
-        }
-    },
-]
-```
-
-In this example:
--  The `test_name` attributes is a unique identifier for the test. In `latency-tests.json`, it must start with `latency_`.
--  The `parameters` attribute control the command line arguments to be used for `benchmark_latency.py`. Note that please use underline `_` instead of the dash `-` when specifying the command line arguments, and `run-performance-benchmarks.sh` will convert the underline to dash when feeding the arguments to `benchmark_latency.py`. For example, the corresponding command line arguments for `benchmark_latency.py` will be `--model meta-llama/Meta-Llama-3-8B --tensor-parallel-size 1 --load-format dummy --num-iters-warmup 5 --num-iters 15`
-
-Note that the performance numbers are highly sensitive to the value of the parameters. Please make sure the parameters are set correctly.
-
-WARNING: The benchmarking script will save json results by itself, so please do not configure `--output-json` parameter in the json file.
-
-
-#### Throughput test
-The tests are specified in `throughput-tests.json`. The syntax is similar to `latency-tests.json`, except for that the parameters will be fed forward to `benchmark_throughput.py`.
-
-The number of this test is also stable -- a slight change on the value of this number might vary the performance numbers by a lot.
-
-#### Serving test
-We test the throughput by using `benchmark_serving.py` with request rate = inf to cover the online serving overhead. The corresponding parameters are in `serving-tests.json`, and here is an example:
-
-```
-[
-    {
-        "test_name": "serving_llama8B_tp1_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "server_parameters": {
-            "model": "meta-llama/Meta-Llama-3-8B",
-            "tensor_parallel_size": 1,
-            "swap_space": 16,
-            "disable_log_stats": "",
-            "disable_log_requests": "",
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Meta-Llama-3-8B",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-]
-```
-
-Inside this example:
-- The `test_name` attribute is also a unique identifier for the test. It must start with `serving_`.
-- The `server-parameters` includes the command line arguments for vLLM server.
-- The `client-parameters` includes the command line arguments for `benchmark_serving.py`.
-- The `qps_list` controls the list of qps for test. It will be used to configure the `--request-rate` parameter in `benchmark_serving.py`
-
-The number of this test is less stable compared to the delay and latency benchmarks (due to randomized sharegpt dataset sampling inside `benchmark_serving.py`), but a large change on this number (e.g. 5% change) still vary the output greatly.
-
-WARNING: The benchmarking script will save json results by itself, so please do not configure `--save-results` or other results-saving-related parameters in `serving-tests.json`.
-
-#### Visualizing the results
-The `convert-results-json-to-markdown.py` helps you put the benchmarking results inside a markdown table, by formatting [descriptions.md](tests/descriptions.md) with real benchmarking results.
-You can find the result presented as a table inside the `buildkite/performance-benchmark` job page.
-If you do not see the table, please wait till the benchmark finish running.
-The json version of the table (together with the json version of the benchmark) will be also attached to the markdown file.
-The raw benchmarking results (in the format of json files) are in the `Artifacts` tab of the benchmarking.
-
-
-
-## Nightly test details
-
-See [nightly-descriptions.md](nightly-descriptions.md) for the detailed description on test workload, models and docker containers of benchmarking other llm engines.
-
-
-#### Workflow
-
-- The [nightly-pipeline.yaml](nightly-pipeline.yaml) specifies the docker containers for different LLM serving engines. 
-- Inside each container, we run [run-nightly-suite.sh](run-nightly-suite.sh), which will probe the serving engine of the current container.
-- The `run-nightly-suite.sh` will redirect the request to `tests/run-[llm serving engine name]-nightly.sh`, which parses the workload described in [nightly-tests.json](tests/nightly-tests.json) and performs the benchmark.
-- At last, we run [scripts/plot-nightly-results.py](scripts/plot-nightly-results.py) to collect and plot the final benchmarking results, and update the results to buildkite.
-
-#### Nightly tests
-
-In [nightly-tests.json](tests/nightly-tests.json), we include the command line arguments for benchmarking commands, together with the benchmarking test cases. The format is highly similar to performance benchmark.
-
-#### Docker containers
-
-The docker containers for benchmarking are specified in `nightly-pipeline.yaml`.
-
-WARNING: the docker versions are HARD-CODED and SHOULD BE ALIGNED WITH `nightly-descriptions.md`. The docker versions need to be hard-coded as there are several version-specific bug fixes inside `tests/run-[llm serving engine name]-nightly.sh`.
-
-WARNING: populating `trt-llm` to latest version is not easy, as it requires updating several protobuf files in [tensorrt-demo](https://github.com/neuralmagic/tensorrt-demo.git).
-

.buildkite/nightly-benchmarks/benchmark-pipeline.yaml

@@ -1,60 +0,0 @@
-steps:
-  - label: "Wait for container to be ready"
-    agents:
-      queue: A100
-    plugins:
-    - kubernetes:
-        podSpec:
-          containers:
-          - image: badouralix/curl-jq
-            command:
-            - sh .buildkite/nightly-benchmarks/scripts/wait-for-image.sh
-  - wait
-  - label: "A100"
-    agents:
-      queue: A100
-    plugins:
-    - kubernetes:
-        podSpec:
-          priorityClassName: perf-benchmark
-          containers:
-          - image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-            command:
-            - bash .buildkite/nightly-benchmarks/scripts/run-performance-benchmarks.sh
-            resources:
-              limits:
-                nvidia.com/gpu: 8
-            volumeMounts:
-            - name: devshm
-              mountPath: /dev/shm
-            env:
-            - name: VLLM_USAGE_SOURCE
-              value: ci-test
-            - name: HF_TOKEN
-              valueFrom:
-                secretKeyRef:
-                  name: hf-token-secret
-                  key: token
-          nodeSelector:
-            nvidia.com/gpu.product: NVIDIA-A100-SXM4-80GB
-          volumes:
-          - name: devshm
-            emptyDir:
-              medium: Memory
-  # - label: "H100"
-  #   agents:
-  #     queue: H100
-  #   plugins:
-  #   - docker#v5.11.0:
-  #       image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
-  #       command:
-  #       - bash
-  #       - .buildkite/nightly-benchmarks/run-benchmarks-suite.sh
-  #       mount-buildkite-agent: true
-  #       propagate-environment: true
-  #       ipc: host
-  #       gpus: all
-  #       environment:
-  #       - VLLM_USAGE_SOURCE
-  #       - HF_TOKEN
-

.buildkite/nightly-benchmarks/nightly-descriptions.md

@@ -1,45 +0,0 @@
-
-# Nightly benchmark
-
-The main goal of this benchmarking is two-fold:
-- Performance clarity: Provide clarity on which one (vllm, tensorrt-llm, lmdeploy and tgi) leads in performance in what workload.
-- Reproducible: one can run the exact same set of benchmarking commands inside the exact same docker by following reproducing instructions in [reproduce.md]().
-
-
-## Docker images
-
-We benchmark vllm, tensorrt-llm, lmdeploy and tgi using the following docker images:
-- vllm/vllm-openai:v0.5.0.post1
-- nvcr.io/nvidia/tritonserver:24.04-trtllm-python-py3
-- openmmlab/lmdeploy:v0.5.0
-- ghcr.io/huggingface/text-generation-inference:2.1
-
-<!-- Please check <a href="artifact://workspace/build/buildkite/vllm/performance-benchmark/.buildkite/nightly-benchmarks/nightly-pipeline.yaml">nightly-pipeline.yaml</a> artifact for more details on how we deploy the docker images. -->
-
-
-## Hardware
-
-One AWS node with 8x NVIDIA A100 GPUs.
-
-
-## Workload description
-
-We benchmark vllm, tensorrt-llm, lmdeploy and tgi using the following workload:
-
-- Input length: randomly sample 500 prompts from ShareGPT dataset (with fixed random seed).
-- Output length: the corresponding output length of these 500 prompts.
-- Models: llama-3 8B, llama-3 70B, mixtral 8x7B.
-- Average QPS (query per second): 4 for the small model (llama-3 8B) and 2 for other two models. For each QPS, the arrival time of each query is determined using a random Poisson process (with fixed random seed).
-- Evaluation metrics: Throughput (higher the better), TTFT (time to the first token, lower the better), ITL (inter-token latency, lower the better).
-
-<!-- Check <a href="artifact://workspace/build/buildkite/vllm/performance-benchmark/.buildkite/nightly-benchmarks/tests/nightly-tests.json">nightly-tests.json</a> artifact for more details. -->
-
-## Plots
-
-In the following plots, the dot shows the mean and the error bar shows the standard error of the mean. Value 0 means that the corresponding benchmark crashed.
-
-<img src="artifact://nightly_results.png" alt="Benchmarking results" height=250 >
-
-## Results
-
-{nightly_results_benchmarking_table}

.buildkite/nightly-benchmarks/nightly-pipeline.yaml

@@ -1,120 +0,0 @@
-common_pod_spec: &common_pod_spec
-  priorityClassName: perf-benchmark
-  nodeSelector:
-    nvidia.com/gpu.product: NVIDIA-A100-SXM4-80GB
-  volumes:
-    - name: devshm
-      emptyDir:
-        medium: Memory
-    - name: hf-cache
-      hostPath:
-        path: /root/.cache/huggingface
-        type: Directory
-
-common_container_settings: &common_container_settings
-  command:
-    - bash .buildkite/nightly-benchmarks/run-nightly-suite.sh
-  resources:
-    limits:
-      nvidia.com/gpu: 8
-  volumeMounts:
-    - name: devshm
-      mountPath: /dev/shm
-    - name: hf-cache
-      mountPath: /root/.cache/huggingface
-  env:
-    - name: VLLM_USAGE_SOURCE
-      value: ci-test
-    - name: HF_HOME
-      value: /root/.cache/huggingface
-    - name: VLLM_SOURCE_CODE_LOC
-      value: /workspace/build/buildkite/vllm/performance-benchmark
-    - name: HF_TOKEN
-      valueFrom:
-        secretKeyRef:
-          name: hf-token-secret
-          key: token
-
-steps:
-  - block: ":rocket: Ready for comparing vllm against alternatives? This will take 4 hours."
-  - label: "A100 trt benchmark"
-    priority: 100
-    agents:
-      queue: A100
-    plugins:
-      - kubernetes:
-          podSpec:
-            <<: *common_pod_spec
-            containers:
-              - image: nvcr.io/nvidia/tritonserver:24.04-trtllm-python-py3
-                <<: *common_container_settings
-
-  - label: "A100 lmdeploy benchmark"
-    priority: 100
-    agents:
-      queue: A100
-    plugins:
-      - kubernetes:
-          podSpec:
-            <<: *common_pod_spec
-            containers:
-              - image: openmmlab/lmdeploy:v0.5.0
-                <<: *common_container_settings
-  
-
-  - label: "A100 vllm benchmark"
-    priority: 100
-    agents:
-      queue: A100
-    plugins:
-      - kubernetes:
-          podSpec:
-            <<: *common_pod_spec
-            containers:
-              - image: vllm/vllm-openai:latest 
-                <<: *common_container_settings
-
-  - label: "A100 tgi benchmark"
-    priority: 100
-    agents:
-      queue: A100
-    plugins:
-      - kubernetes:
-          podSpec:
-            <<: *common_pod_spec
-            containers:
-              - image: ghcr.io/huggingface/text-generation-inference:2.1 
-                <<: *common_container_settings
-        
-  - wait
-
-  - label: "Plot"
-    priority: 100
-    agents:
-      queue: A100
-    plugins:
-      - kubernetes:
-          podSpec:
-            <<: *common_pod_spec
-            containers:
-            - image: vllm/vllm-openai:v0.5.0.post1
-              command:
-              - bash .buildkite/nightly-benchmarks/scripts/nightly-annotate.sh
-              resources:
-                limits:
-                  nvidia.com/gpu: 8
-              volumeMounts:
-              - name: devshm
-                mountPath: /dev/shm
-              env:
-              - name: VLLM_USAGE_SOURCE
-                value: ci-test
-              - name: VLLM_SOURCE_CODE_LOC
-                value: /workspace/build/buildkite/vllm/performance-benchmark
-              - name: HF_TOKEN
-                valueFrom:
-                  secretKeyRef:
-                    name: hf-token-secret
-                    key: token
-
-  - wait

.buildkite/nightly-benchmarks/performance-benchmarks-descriptions.md

@@ -1,62 +0,0 @@
-
-## Latency tests
-
-- Input length: 32 tokens.
-- Output length: 128 tokens.
-- Batch size: fixed (8).
-- Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
-- Evaluation metrics: end-to-end latency (mean, median, p99).
-
-
-{latency_tests_markdown_table}
-
-
-## Throughput tests
-
-- Input length: randomly sample 200 prompts from ShareGPT dataset (with fixed random seed).
-- Output length: the corresponding output length of these 200 prompts.
-- Batch size: dynamically determined by vllm to achieve maximum throughput.
-- Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
-- Evaluation metrics: throughput.
-
-
-{throughput_tests_markdown_table}
-
-
-## Serving tests
-
-- Input length: randomly sample 200 prompts from ShareGPT dataset (with fixed random seed).
-- Output length: the corresponding output length of these 200 prompts.
-- Batch size: dynamically determined by vllm and the arrival pattern of the requests.
-- **Average QPS (query per second)**: 1, 4, 16 and inf. QPS = inf means all requests come at once. For other QPS values, the arrival time of each query is determined using a random Poisson process (with fixed random seed).
-- Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
-- We also added a speculative decoding test for llama-3 70B, under QPS 2
-- Evaluation metrics: throughput, TTFT (time to the first token, with mean, median and p99), ITL (inter-token latency, with mean, median and p99).
-
-
-{serving_tests_markdown_table}
-
-
-## json version of the benchmarking tables
-
-This section contains the data of the markdown tables above in JSON format. 
-You can load the benchmarking tables into pandas dataframes as follows:
-
-```python
-import json
-import pandas as pd
-
-benchmarking_results_json = """The json string"""
-benchmarking_results = json.loads(benchmarking_results_json)
-latency_results = pd.DataFrame.from_dict(benchmarking_results["latency"])
-throughput_results = pd.DataFrame.from_dict(benchmarking_results["throughput"])
-serving_results = pd.DataFrame.from_dict(benchmarking_results["serving"])
-```
-
-The json string for all benchmarking tables:
-```json
-{benchmarking_results_in_json_string}
-```
-
-You can also check the raw experiment data in the Artifact tab of the Buildkite page.
-

.buildkite/nightly-benchmarks/run-nightly-suite.sh

@@ -1,76 +0,0 @@
-#!/bin/bash
-
-set -o pipefail
-set -x
-
-check_gpus() {
-    # check the number of GPUs and GPU type.
-    declare -g gpu_count=$(nvidia-smi --list-gpus | wc -l)
-    if [[ $gpu_count -gt 0 ]]; then
-        echo "GPU found."
-    else
-        echo "Need at least 1 GPU to run benchmarking."
-        exit 1
-    fi
-    declare -g gpu_type=$(echo $(nvidia-smi --query-gpu=name --format=csv,noheader) | awk '{print $2}')
-    echo "GPU type is $gpu_type"
-}
-
-check_hf_token() {
-    # check if HF_TOKEN is available and valid
-    if [[ -z "$HF_TOKEN" ]]; then
-        echo "Error: HF_TOKEN is not set."
-        exit 1
-    elif [[ ! "$HF_TOKEN" =~ ^hf_ ]]; then
-        echo "Error: HF_TOKEN does not start with 'hf_'."
-        exit 1
-    else
-        echo "HF_TOKEN is set and valid."
-    fi
-}
-
-main() {
-
-    check_gpus
-    check_hf_token
-
-    df -h
-
-    (which wget && which curl) || (apt-get update && apt-get install -y wget curl)
-    (which jq) || (apt-get update && apt-get -y install jq)
-
-    cd $VLLM_SOURCE_CODE_LOC/benchmarks
-    wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json
-    
-
-    # run lmdeploy
-    if which lmdeploy >/dev/null; then
-        echo "lmdeploy is available, redirect to run-lmdeploy-nightly.sh"
-        bash ../.buildkite/nightly-benchmarks/scripts/run-lmdeploy-nightly.sh
-        exit 0
-    fi
-
-    # run tgi
-    if [ -e /tgi-entrypoint.sh ]; then
-        echo "tgi is available, redirect to run-tgi-nightly.sh"
-        bash ../.buildkite/nightly-benchmarks/scripts/run-tgi-nightly.sh
-        exit 0
-    fi
-
-    # run trt
-    if which trtllm-build >/dev/null; then
-        echo "trtllm is available, redirect to run-trt-nightly.sh"
-        bash ../.buildkite/nightly-benchmarks/scripts/run-trt-nightly.sh
-        exit 0
-    fi
-
-    # run vllm
-    if [ -e /vllm-workspace ]; then
-        echo "vllm is available, redirect to run-vllm-nightly.sh"
-        bash ../.buildkite/nightly-benchmarks/scripts/run-vllm-nightly.sh
-        exit 0
-    fi
-
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/convert-results-json-to-markdown.py

@@ -1,192 +0,0 @@
-import json
-import os
-from pathlib import Path
-
-import pandas as pd
-from tabulate import tabulate
-
-results_folder = Path("results/")
-
-# latency results and the keys that will be printed into markdown
-latency_results = []
-latency_column_mapping = {
-    "test_name": "Test name",
-    "gpu_type": "GPU",
-    "avg_latency": "Mean latency (ms)",
-    # "P10": "P10 (s)",
-    # "P25": "P25 (s)",
-    "P50": "Median latency (ms)",
-    # "P75": "P75 (s)",
-    # "P90": "P90 (s)",
-    "P99": "P99 latency (ms)",
-}
-
-# throughput tests and the keys that will be printed into markdown
-throughput_results = []
-throughput_results_column_mapping = {
-    "test_name": "Test name",
-    "gpu_type": "GPU",
-    # "num_requests": "# of req.",
-    # "total_num_tokens": "Total # of tokens",
-    # "elapsed_time": "Elapsed time (s)",
-    "requests_per_second": "Tput (req/s)",
-    # "tokens_per_second": "Tput (tok/s)",
-}
-
-# serving results and the keys that will be printed into markdown
-serving_results = []
-serving_column_mapping = {
-    "test_name": "Test name",
-    "gpu_type": "GPU",
-    # "completed": "# of req.",
-    "request_throughput": "Tput (req/s)",
-    # "input_throughput": "Input Tput (tok/s)",
-    # "output_throughput": "Output Tput (tok/s)",
-    "mean_ttft_ms": "Mean TTFT (ms)",
-    "median_ttft_ms": "Median TTFT (ms)",
-    "p99_ttft_ms": "P99 TTFT (ms)",
-    # "mean_tpot_ms": "Mean TPOT (ms)",
-    # "median_tpot_ms": "Median",
-    # "p99_tpot_ms": "P99",
-    "mean_itl_ms": "Mean ITL (ms)",
-    "median_itl_ms": "Median ITL (ms)",
-    "p99_itl_ms": "P99 ITL (ms)",
-}
-
-
-def read_markdown(file):
-    if os.path.exists(file):
-        with open(file, "r") as f:
-            return f.read() + "\n"
-    else:
-        return f"{file} not found.\n"
-
-
-def results_to_json(latency, throughput, serving):
-    return json.dumps({
-        'latency': latency.to_dict(),
-        'throughput': throughput.to_dict(),
-        'serving': serving.to_dict()
-    })
-
-
-if __name__ == "__main__":
-
-    # collect results
-    for test_file in results_folder.glob("*.json"):
-
-        with open(test_file, "r") as f:
-            raw_result = json.loads(f.read())
-
-        if "serving" in str(test_file):
-            # this result is generated via `benchmark_serving.py`
-
-            # attach the benchmarking command to raw_result
-            with open(test_file.with_suffix(".commands"), "r") as f:
-                command = json.loads(f.read())
-            raw_result.update(command)
-
-            # update the test name of this result
-            raw_result.update({"test_name": test_file.stem})
-
-            # add the result to raw_result
-            serving_results.append(raw_result)
-            continue
-
-        elif "latency" in f.name:
-            # this result is generated via `benchmark_latency.py`
-
-            # attach the benchmarking command to raw_result
-            with open(test_file.with_suffix(".commands"), "r") as f:
-                command = json.loads(f.read())
-            raw_result.update(command)
-
-            # update the test name of this result
-            raw_result.update({"test_name": test_file.stem})
-
-            # get different percentiles
-            for perc in [10, 25, 50, 75, 90, 99]:
-                # Multiply 1000 to convert the time unit from s to ms
-                raw_result.update(
-                    {f"P{perc}": 1000 * raw_result["percentiles"][str(perc)]})
-            raw_result["avg_latency"] = raw_result["avg_latency"] * 1000
-
-            # add the result to raw_result
-            latency_results.append(raw_result)
-            continue
-
-        elif "throughput" in f.name:
-            # this result is generated via `benchmark_throughput.py`
-
-            # attach the benchmarking command to raw_result
-            with open(test_file.with_suffix(".commands"), "r") as f:
-                command = json.loads(f.read())
-            raw_result.update(command)
-
-            # update the test name of this result
-            raw_result.update({"test_name": test_file.stem})
-
-            # add the result to raw_result
-            throughput_results.append(raw_result)
-            continue
-
-        print(f"Skipping {test_file}")
-
-    latency_results = pd.DataFrame.from_dict(latency_results)
-    serving_results = pd.DataFrame.from_dict(serving_results)
-    throughput_results = pd.DataFrame.from_dict(throughput_results)
-
-    raw_results_json = results_to_json(latency_results, throughput_results,
-                                       serving_results)
-
-    # remapping the key, for visualization purpose
-    if not latency_results.empty:
-        latency_results = latency_results[list(
-            latency_column_mapping.keys())].rename(
-                columns=latency_column_mapping)
-    if not serving_results.empty:
-        serving_results = serving_results[list(
-            serving_column_mapping.keys())].rename(
-                columns=serving_column_mapping)
-    if not throughput_results.empty:
-        throughput_results = throughput_results[list(
-            throughput_results_column_mapping.keys())].rename(
-                columns=throughput_results_column_mapping)
-
-    processed_results_json = results_to_json(latency_results,
-                                             throughput_results,
-                                             serving_results)
-
-    # get markdown tables
-    latency_md_table = tabulate(latency_results,
-                                headers='keys',
-                                tablefmt='pipe',
-                                showindex=False)
-    serving_md_table = tabulate(serving_results,
-                                headers='keys',
-                                tablefmt='pipe',
-                                showindex=False)
-    throughput_md_table = tabulate(throughput_results,
-                                   headers='keys',
-                                   tablefmt='pipe',
-                                   showindex=False)
-
-    # document the result
-    with open(results_folder / "benchmark_results.md", "w") as f:
-
-        results = read_markdown("../.buildkite/nightly-benchmarks/" +
-                                "performance-benchmarks-descriptions.md")
-        results = results.format(
-            latency_tests_markdown_table=latency_md_table,
-            throughput_tests_markdown_table=throughput_md_table,
-            serving_tests_markdown_table=serving_md_table,
-            benchmarking_results_in_json_string=processed_results_json)
-        f.write(results)
-
-    # document benchmarking results in json
-    with open(results_folder / "benchmark_results.json", "w") as f:
-
-        results = latency_results.to_dict(
-            orient='records') + throughput_results.to_dict(
-                orient='records') + serving_results.to_dict(orient='records')
-        f.write(json.dumps(results))

.buildkite/nightly-benchmarks/scripts/download-tokenizer.py

@@ -1,26 +0,0 @@
-import argparse
-
-from transformers import AutoTokenizer
-
-
-def main(model, cachedir):
-    # Load the tokenizer and save it to the specified directory
-    tokenizer = AutoTokenizer.from_pretrained(model)
-    tokenizer.save_pretrained(cachedir)
-    print(f"Tokenizer saved to {cachedir}")
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(
-        description="Download and save Hugging Face tokenizer")
-    parser.add_argument("--model",
-                        type=str,
-                        required=True,
-                        help="Name of the model")
-    parser.add_argument("--cachedir",
-                        type=str,
-                        required=True,
-                        help="Directory to save the tokenizer")
-
-    args = parser.parse_args()
-    main(args.model, args.cachedir)

.buildkite/nightly-benchmarks/scripts/get-lmdeploy-modelname.py

@@ -1,6 +0,0 @@
-from lmdeploy.serve.openai.api_client import APIClient
-
-api_client = APIClient("http://localhost:8000")
-model_name = api_client.available_models[0]
-
-print(model_name)

.buildkite/nightly-benchmarks/scripts/launch-trt-server.sh

@@ -1,102 +0,0 @@
-#!/bin/bash
-
-
-server_params=$1
-common_params=$2
-
-
-
-model_path=$(echo "$common_params" | jq -r '.model')
-model_name="${model_path#*/}"
-model_type=$(echo "$server_params" | jq -r '.model_type')
-model_dtype=$(echo "$server_params" | jq -r '.model_dtype')
-model_tp_size=$(echo "$common_params" | jq -r '.tp')
-max_batch_size=$(echo "$server_params" | jq -r '.max_batch_size')
-max_input_len=$(echo "$server_params" | jq -r '.max_input_len')
-max_output_len=$(echo "$server_params" | jq -r '.max_output_len')
-trt_llm_version=$(echo "$server_params" | jq -r '.trt_llm_version')
-
-cd ~
-rm -rf models
-mkdir -p models
-cd models
-models_dir=$(pwd)
-trt_model_path=${models_dir}/${model_name}-trt-ckpt
-trt_engine_path=${models_dir}/${model_name}-trt-engine
-
-cd ~
-rm -rf tensorrt-demo
-git clone https://github.com/neuralmagic/tensorrt-demo.git
-cd tensorrt-demo
-tensorrt_demo_dir=$(pwd)
-
-# make sure the parameter inside tensorrt_demo is consistent to envvar
-sed -i.bak "/key: \"tokenizer_dir\"/,/string_value:/s|string_value: \".*\"|string_value: \"$model_path\"|" ./triton_model_repo/postprocessing/config.pbtxt
-sed -i.bak "/key: \"tokenizer_dir\"/,/string_value:/s|string_value: \".*\"|string_value: \"$model_path\"|" ./triton_model_repo/preprocessing/config.pbtxt
-sed -i.bak "s|\(max_batch_size:\s*\)[0-9]*|\1$max_batch_size|g" ./triton_model_repo/ensemble/config.pbtxt
-sed -i.bak "s|\(max_batch_size:\s*\)[0-9]*|\1$max_batch_size|g" ./triton_model_repo/preprocessing/config.pbtxt
-sed -i.bak "s|\(max_batch_size:\s*\)[0-9]*|\1$max_batch_size|g" ./triton_model_repo/postprocessing/config.pbtxt
-sed -i.bak "s|\(max_batch_size:\s*\)[0-9]*|\1$max_batch_size|g" ./triton_model_repo/tensorrt_llm_bls/config.pbtxt
-
-
-cd /
-rm -rf tensorrtllm_backend
-git clone https://github.com/triton-inference-server/tensorrtllm_backend.git
-git lfs install
-cd tensorrtllm_backend
-git checkout $trt_llm_version
-tensorrtllm_backend_dir=$(pwd)
-git submodule update --init --recursive
-cp -r ${tensorrt_demo_dir}/triton_model_repo ${tensorrtllm_backend_dir}/
-
-cd /tensorrtllm_backend
-cd ./tensorrt_llm/examples/${model_type}
-
-
-if echo "$common_params" | jq -e 'has("fp8")' > /dev/null; then
-
-    echo "Key 'fp8' exists in common params. Use quantize.py instead of convert_checkpoint.py"
-    echo "Reference: https://github.com/NVIDIA/TensorRT-LLM/blob/main/examples/llama/README.md"
-    python ../quantization/quantize.py \
-        --model_dir ${model_path} \
-        --dtype ${model_dtype} \
-        --tp_size ${model_tp_size} \
-        --output_dir ${trt_model_path} \
-        --qformat fp8 \
-        --kv_cache_dtype fp8 \
-        --calib_size 2
-
-else
-
-    echo "Key 'fp8' does not exist in common params. Use convert_checkpoint.py"
-    python3 convert_checkpoint.py \
-        --model_dir ${model_path} \
-        --dtype ${model_dtype} \
-        --tp_size ${model_tp_size} \
-        --output_dir ${trt_model_path}
-
-fi
-
-
-
-trtllm-build \
---checkpoint_dir=${trt_model_path} \
---gpt_attention_plugin=${model_dtype} \
---gemm_plugin=${model_dtype} \
---remove_input_padding=enable \
---paged_kv_cache=enable \
---tp_size=${model_tp_size} \
---max_batch_size=${max_batch_size} \
---max_input_len=${max_input_len} \
---max_output_len=${max_output_len} \
---max_num_tokens=${max_output_len} \
---opt_num_tokens=${max_output_len} \
---output_dir=${trt_engine_path} 
-
-cd /tensorrtllm_backend/triton_model_repo
-rm -rf ./tensorrt_llm/1/*
-cp -r ${trt_engine_path}/* ./tensorrt_llm/1
-cd /tensorrtllm_backend
-python3 scripts/launch_triton_server.py \
---world_size=${model_tp_size} \
---model_repo=/tensorrtllm_backend/triton_model_repo &

.buildkite/nightly-benchmarks/scripts/nightly-annotate.sh

@@ -1,40 +0,0 @@
-#!/bin/bash
-
-set -ex
-set -o pipefail
-
-
-main() {
-
-    (which wget && which curl) || (apt-get update && apt-get install -y wget curl)
-    (which jq) || (apt-get update && apt-get -y install jq)
-
-    if [ ! -f /workspace/buildkite-agent ]; then
-        echo "buildkite-agent binary not found. Skip plotting the results."
-        exit 0
-    fi
-
-    # initial annotation
-    description="$VLLM_SOURCE_CODE_LOC/.buildkite/nightly-benchmarks/nightly-descriptions.md"
-
-    # download results
-    cd $VLLM_SOURCE_CODE_LOC/benchmarks
-    mkdir -p results/
-    /workspace/buildkite-agent artifact download 'results/*nightly_results.json' results/
-    ls
-    ls results/
-
-    # generate figures
-    python3 -m pip install tabulate pandas matplotlib
-    python3 $VLLM_SOURCE_CODE_LOC/.buildkite/nightly-benchmarks/scripts/plot-nightly-results.py \
-        --description $description \
-        --results-folder results/
-    
-    # upload results and figures
-    /workspace/buildkite-agent artifact upload "nightly_results.png"
-    /workspace/buildkite-agent artifact upload $VLLM_SOURCE_CODE_LOC/.buildkite/nightly-benchmarks/nightly-pipeline.yaml
-    /workspace/buildkite-agent artifact upload $VLLM_SOURCE_CODE_LOC/.buildkite/nightly-benchmarks/tests/nightly-tests.json
-    /workspace/buildkite-agent annotate --style "success" --context "nightly-benchmarks-results" --append < nightly_results.md
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/plot-nightly-results.py

@@ -1,135 +0,0 @@
-import argparse
-import json
-import math
-from pathlib import Path
-
-import matplotlib.pyplot as plt
-import pandas as pd
-from tabulate import tabulate
-
-
-def parse_arguments():
-    parser = argparse.ArgumentParser(
-        description=
-        'Parse command line arguments for summary-nightly-results script.')
-    parser.add_argument('--results-folder',
-                        type=str,
-                        required=True,
-                        help='The folder where the results are stored.')
-    parser.add_argument('--description',
-                        type=str,
-                        required=True,
-                        help='Description of the results.')
-
-    args = parser.parse_args()
-    return args
-
-
-def main(args):
-    bar_colors = ['#56B4E9', '#009E73', '#D55E00', '#E69F00']
-    results_folder = Path(args.results_folder)
-
-    results = []
-
-    # collect results
-    for test_file in results_folder.glob("*_nightly_results.json"):
-        with open(test_file, "r") as f:
-            results = results + json.loads(f.read())
-
-    # generate markdown table
-    df = pd.DataFrame.from_dict(results)
-
-    md_table = tabulate(df, headers='keys', tablefmt='pipe', showindex=False)
-
-    with open(args.description, "r") as f:
-        description = f.read()
-
-    description = description.format(
-        nightly_results_benchmarking_table=md_table)
-
-    with open("nightly_results.md", "w") as f:
-        f.write(description)
-
-    plt.rcParams.update({'font.size': 20})
-
-    # plot results
-    fig, axes = plt.subplots(3, 3, figsize=(16, 14))
-    fig.subplots_adjust(hspace=1)
-    methods = ["vllm", "trt", "lmdeploy", "tgi"]
-    for i, model in enumerate(["llama8B", "llama70B", "mixtral8x7B"]):
-        for j, metric in enumerate(["TTFT", "ITL"]):
-            means, stds = [], []
-            for method in methods:
-                target = df['Test name'].str.contains(model)
-                target = target & df['Engine'].str.contains(method)
-                filtered_df = df[target]
-
-                if filtered_df.empty:
-                    means.append(0.)
-                    stds.append(0.)
-                else:
-                    means.append(filtered_df[f"Mean {metric} (ms)"].values[0])
-                    std = filtered_df[f"Std {metric} (ms)"].values[0]
-                    success = filtered_df["Successful req."].values[0]
-                    stds.append(std / math.sqrt(success))
-
-            print(model, metric)
-            print(means, stds)
-
-            ax = axes[i, j + 1]
-
-            bars = ax.bar(
-                ["vllm", "trt", "lmdeploy", "tgi"],
-                means,
-                yerr=stds,
-                capsize=10,
-            )
-            for idx, bar in enumerate(bars):
-                bar.set_color(bar_colors[idx])
-            ax.set_ylim(bottom=0)
-
-            ax.set_ylabel(f"{metric} (ms)")
-            ax.set_title(f"{model} {metric}")
-            ax.grid(axis='y')
-
-        metric = "Tput"
-        j = 0
-        if True:
-            tputs = []
-            for method in methods:
-                target = df['Test name'].str.contains(model)
-                target = target & df['Engine'].str.contains(method)
-                filtered_df = df[target]
-
-                if filtered_df.empty:
-                    tputs.append(0.)
-                else:
-                    input_tput = filtered_df["Input Tput (tok/s)"].values[0]
-                    output_tput = filtered_df["Output Tput (tok/s)"].values[0]
-                    tputs.append(input_tput + output_tput)
-
-            print(model, metric)
-            print(tputs)
-
-            ax = axes[i, j]
-
-            bars = ax.bar(
-                ["vllm", "trt", "lmdeploy", "tgi"],
-                tputs,
-            )
-            for idx, bar in enumerate(bars):
-                bar.set_color(bar_colors[idx])
-
-            ax.set_ylim(bottom=0)
-
-            ax.set_ylabel("Tput (token/s)")
-            ax.set_title(f"{model} {metric}")
-            ax.grid(axis='y')
-
-    fig.tight_layout()
-    fig.savefig("nightly_results.png", bbox_inches='tight', dpi=400)
-
-
-if __name__ == '__main__':
-    args = parse_arguments()
-    main(args)

.buildkite/nightly-benchmarks/scripts/run-lmdeploy-nightly.sh

@@ -1,218 +0,0 @@
-#!/bin/bash
-
-set -o pipefail
-
-check_gpus() {
-  # check the number of GPUs and GPU type.
-  declare -g gpu_count=$(nvidia-smi --list-gpus | wc -l)
-  if [[ $gpu_count -gt 0 ]]; then
-    echo "GPU found."
-  else
-    echo "Need at least 1 GPU to run benchmarking."
-    exit 1
-  fi
-  declare -g gpu_type=$(echo $(nvidia-smi --query-gpu=name --format=csv,noheader) | awk '{print $2}')
-  echo "GPU type is $gpu_type"
-}
-
-kill_gpu_processes() {
-  pkill lmdeploy || true
-  # waiting for GPU processes to be fully killed
-  sleep 10
-  # Print the GPU memory usage
-  # so that we know if all GPU processes are killed.
-  gpu_memory_usage=$(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits -i 0)
-  # The memory usage should be 0 MB.
-  echo "GPU 0 Memory Usage: $gpu_memory_usage MB"
-}
-
-json2args() {
-  # transforms the JSON string to command line args, and '_' is replaced to '-'
-  # example:
-  # input: { "model": "meta-llama/Llama-2-7b-chat-hf", "tensor_parallel_size": 1 }
-  # output: --model meta-llama/Llama-2-7b-chat-hf --tensor-parallel-size 1
-  local json_string=$1
-  local args=$(
-    echo "$json_string" | jq -r '
-      to_entries |
-      map("--" + (.key | gsub("_"; "-")) + " " + (.value | tostring)) |
-      join(" ")
-    '
-  )
-  echo "$args"
-}
-
-wait_for_server() {
-  # wait for vllm server to start
-  # return 1 if vllm server crashes
-  timeout 1200 bash -c '
-    until curl -s localhost:8000/v1/completions > /dev/null; do
-      sleep 1
-    done' && return 0 || return 1
-}
-
-run_serving_tests() {
-  # run serving tests using `benchmark_serving.py`
-  # $1: a json file specifying serving test cases
-
-  local serving_test_file
-  serving_test_file=$1
-
-  # Iterate over serving tests
-  jq -c '.[]' "$serving_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-    
-    # append lmdeploy to the test name
-    test_name=lmdeploy_$test_name
-    
-    # get common parameters
-    common_params=$(echo "$params" | jq -r '.common_parameters')
-    model=$(echo "$common_params" | jq -r '.model')
-    tp=$(echo "$common_params" | jq -r '.tp')
-    dataset_name=$(echo "$common_params" | jq -r '.dataset_name')
-    dataset_path=$(echo "$common_params" | jq -r '.dataset_path')
-    port=$(echo "$common_params" | jq -r '.port')
-    num_prompts=$(echo "$common_params" | jq -r '.num_prompts')
-
-    
-
-    # get client and server arguments
-    server_params=$(echo "$params" | jq -r '.lmdeploy_server_parameters')
-    client_params=$(echo "$params" | jq -r '.lmdeploy_client_parameters')
-    server_args=$(json2args "$server_params")
-    client_args=$(json2args "$client_params")
-    qps_list=$(echo "$params" | jq -r '.qps_list')
-    qps_list=$(echo "$qps_list" | jq -r '.[] | @sh')
-    echo "Running over qps list $qps_list"
-
-    # check if there is enough GPU to run the test
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
-      continue
-    fi
-
-    # prepare tokenizer
-    rm -rf /tokenizer_cache
-    mkdir /tokenizer_cache
-    python ../.buildkite/nightly-benchmarks/scripts/download-tokenizer.py \
-      --model "$model" \
-      --cachedir /tokenizer_cache
-
-    server_command="lmdeploy serve api_server $model \
-      --tp $tp \
-      --server-port $port \
-      $server_args"
-
-    # run the server
-    echo "Running test case $test_name"
-    echo "Server command: $server_command"
-    bash -c "$server_command" &
-
-    # wait until the server is alive
-    wait_for_server
-    if [ $? -eq 0 ]; then
-      echo ""
-      echo "lmdeploy server is up and running."
-    else
-      echo ""
-      echo "lmdeploy failed to start within the timeout period."
-      break
-    fi
-
-    # get model name
-    model_name=$(python ../.buildkite/nightly-benchmarks/scripts/get-lmdeploy-modelname.py)
-
-    # iterate over different QPS
-    for qps in $qps_list; do
-      # remove the surrounding single quote from qps
-      if [[ "$qps" == *"inf"* ]]; then
-        echo "qps was $qps"
-        qps="inf"
-        echo "now qps is $qps"
-      fi
-
-      new_test_name=$test_name"_qps_"$qps
-
-      client_command="python3 benchmark_serving.py \
-        --backend lmdeploy \
-        --tokenizer /tokenizer_cache \
-        --dataset-name $dataset_name \
-        --dataset-path $dataset_path \
-        --num-prompts $num_prompts \
-        --port $port \
-        --save-result \
-        --result-dir $RESULTS_FOLDER \
-        --result-filename ${new_test_name}.json \
-        --request-rate $qps \
-        --model \"$model_name\" \
-        $client_args"
-
-      echo "Running test case $test_name with qps $qps"
-      echo "Client command: $client_command"
-
-      eval "$client_command"
-
-      # record the benchmarking commands
-      jq_output=$(jq -n \
-        --arg server "$server_command" \
-        --arg client "$client_command" \
-        --arg gpu "$gpu_type" \
-        --arg engine "lmdeploy" \
-        '{
-          server_command: $server,
-          client_command: $client,
-          gpu_type: $gpu,
-          engine: $engine
-        }')
-      echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"
-
-    done
-
-    # clean up
-    kill_gpu_processes
-    rm -rf /root/.cache/huggingface/*
-  done
-}
-
-
-upload_to_buildkite() {
-  # upload the benchmarking results to buildkite
-
-  # if the agent binary is not found, skip uploading the results, exit 0
-  if [ ! -f /workspace/buildkite-agent ]; then
-    echo "buildkite-agent binary not found. Skip uploading the results."
-    return 0
-  fi
-  # /workspace/buildkite-agent annotate --style "success" --context "benchmark-results" --append < $RESULTS_FOLDER/${CURRENT_LLM_SERVING_ENGINE}_nightly_results.md
-  /workspace/buildkite-agent artifact upload "$RESULTS_FOLDER/*"
-}
-
-
-main() {
-
-  check_gpus
-  # enter vllm directory
-  cd $VLLM_SOURCE_CODE_LOC/benchmarks
-
-  declare -g RESULTS_FOLDER=results/
-  mkdir -p $RESULTS_FOLDER
-  BENCHMARK_ROOT=../.buildkite/nightly-benchmarks/
-
-  python -m pip install transformers==4.41.2
-
-  export CURRENT_LLM_SERVING_ENGINE=lmdeploy
-  run_serving_tests $BENCHMARK_ROOT/tests/nightly-tests.json
-  python -m pip install tabulate pandas
-  python $BENCHMARK_ROOT/scripts/summary-nightly-results.py
-  upload_to_buildkite
-
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/run-performance-benchmarks.sh

@@ -1,381 +0,0 @@
-#!/bin/bash
-
-# This script should be run inside the CI process
-# This script assumes that we are already inside the vllm/ directory
-# Benchmarking results will be available inside vllm/benchmarks/results/
-
-# Do not set -e, as the mixtral 8x22B model tends to crash occasionally
-# and we still want to see other benchmarking results even when mixtral crashes.
-set -o pipefail
-
-check_gpus() {
-  # check the number of GPUs and GPU type.
-  declare -g gpu_count=$(nvidia-smi --list-gpus | wc -l)
-  if [[ $gpu_count -gt 0 ]]; then
-    echo "GPU found."
-  else
-    echo "Need at least 1 GPU to run benchmarking."
-    exit 1
-  fi
-  declare -g gpu_type=$(echo $(nvidia-smi --query-gpu=name --format=csv,noheader) | awk '{print $2}')
-  echo "GPU type is $gpu_type"
-}
-
-check_hf_token() {
-  # check if HF_TOKEN is available and valid
-  if [[ -z "$HF_TOKEN" ]]; then
-    echo "Error: HF_TOKEN is not set."
-    exit 1
-  elif [[ ! "$HF_TOKEN" =~ ^hf_ ]]; then
-    echo "Error: HF_TOKEN does not start with 'hf_'."
-    exit 1
-  else
-    echo "HF_TOKEN is set and valid."
-  fi
-}
-
-ensure_sharegpt_downloaded() {
-  local FILE=ShareGPT_V3_unfiltered_cleaned_split.json
-  if [ ! -f "$FILE" ]; then
-    wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/$FILE
-  else
-    echo "$FILE already exists."
-  fi
-}
-
-json2args() {
-  # transforms the JSON string to command line args, and '_' is replaced to '-'
-  # example:
-  # input: { "model": "meta-llama/Llama-2-7b-chat-hf", "tensor_parallel_size": 1 }
-  # output: --model meta-llama/Llama-2-7b-chat-hf --tensor-parallel-size 1
-  local json_string=$1
-  local args=$(
-    echo "$json_string" | jq -r '
-      to_entries |
-      map("--" + (.key | gsub("_"; "-")) + " " + (.value | tostring)) |
-      join(" ")
-    '
-  )
-  echo "$args"
-}
-
-wait_for_server() {
-  # wait for vllm server to start
-  # return 1 if vllm server crashes
-  timeout 1200 bash -c '
-    until curl -X POST localhost:8000/v1/completions; do
-      sleep 1
-    done' && return 0 || return 1
-}
-
-kill_processes_launched_by_current_bash() {
-  # Kill all python processes launched from current bash script
-  current_shell_pid=$$
-  processes=$(ps -eo pid,ppid,command | awk -v ppid="$current_shell_pid" -v proc="$1" '$2 == ppid && $3 ~ proc {print $1}')
-  if [ -n "$processes" ]; then
-    echo "Killing the following processes matching '$1':"
-    echo "$processes"
-    echo "$processes" | xargs kill -9
-  else
-    echo "No processes found matching '$1'."
-  fi
-}
-
-kill_gpu_processes() {
-
-  ps -aux
-  lsof -t -i:8000 | xargs -r kill -9
-  pkill -f pt_main_thread
-  # this line doesn't work now
-  # ps aux | grep python | grep openai | awk '{print $2}' | xargs -r kill -9
-  pkill -f python3
-  pkill -f /usr/bin/python3
-
-
-  # wait until GPU memory usage smaller than 1GB
-  while [ $(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits | head -n 1) -ge 1000 ]; do
-    sleep 1
-  done
-
-  # remove vllm config file
-  rm -rf ~/.config/vllm
-
-}
-
-upload_to_buildkite() {
-  # upload the benchmarking results to buildkite
-
-  # if the agent binary is not found, skip uploading the results, exit 0
-  # Check if buildkite-agent is available in the PATH or at /workspace/buildkite-agent
-  if command -v buildkite-agent >/dev/null 2>&1; then
-    BUILDKITE_AGENT_COMMAND="buildkite-agent"
-  elif [ -f /workspace/buildkite-agent ]; then
-    BUILDKITE_AGENT_COMMAND="/workspace/buildkite-agent"
-  else
-    echo "buildkite-agent binary not found. Skip uploading the results."
-    return 0
-  fi
-
-  # Use the determined command to annotate and upload artifacts
-  $BUILDKITE_AGENT_COMMAND annotate --style "info" --context "$BUILDKITE_LABEL-benchmark-results" <$RESULTS_FOLDER/benchmark_results.md
-  $BUILDKITE_AGENT_COMMAND artifact upload "$RESULTS_FOLDER/*"
-}
-
-run_latency_tests() {
-  # run latency tests using `benchmark_latency.py`
-  # $1: a json file specifying latency test cases
-
-  local latency_test_file
-  latency_test_file=$1
-
-  # Iterate over latency tests
-  jq -c '.[]' "$latency_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-    if [[ ! "$test_name" =~ ^latency_ ]]; then
-      echo "In latency-test.json, test_name must start with \"latency_\"."
-      exit 1
-    fi
-
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-
-    # get arguments
-    latency_params=$(echo "$params" | jq -r '.parameters')
-    latency_args=$(json2args "$latency_params")
-
-    # check if there is enough GPU to run the test
-    tp=$(echo "$latency_params" | jq -r '.tensor_parallel_size')
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $testname."
-      continue
-    fi
-
-    latency_command="python3 benchmark_latency.py \
-      --output-json $RESULTS_FOLDER/${test_name}.json \
-      $latency_args"
-
-    echo "Running test case $test_name"
-    echo "Latency command: $latency_command"
-
-    # recoding benchmarking command ang GPU command
-    jq_output=$(jq -n \
-      --arg latency "$latency_command" \
-      --arg gpu "$gpu_type" \
-      '{
-        latency_command: $latency,
-        gpu_type: $gpu
-      }')
-    echo "$jq_output" >"$RESULTS_FOLDER/$test_name.commands"
-
-    # run the benchmark
-    eval "$latency_command"
-
-    kill_gpu_processes
-
-  done
-}
-
-run_throughput_tests() {
-  # run throughput tests using `benchmark_throughput.py`
-  # $1: a json file specifying throughput test cases
-
-  local throughput_test_file
-  throughput_test_file=$1
-
-  # Iterate over throughput tests
-  jq -c '.[]' "$throughput_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-    if [[ ! "$test_name" =~ ^throughput_ ]]; then
-      echo "In throughput-test.json, test_name must start with \"throughput_\"."
-      exit 1
-    fi
-
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-
-    # get arguments
-    throughput_params=$(echo "$params" | jq -r '.parameters')
-    throughput_args=$(json2args "$throughput_params")
-
-    # check if there is enough GPU to run the test
-    tp=$(echo $throughput_params | jq -r '.tensor_parallel_size')
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $testname."
-      continue
-    fi
-
-    throughput_command="python3 benchmark_throughput.py \
-      --output-json $RESULTS_FOLDER/${test_name}.json \
-      $throughput_args"
-
-    echo "Running test case $test_name"
-    echo "Throughput command: $throughput_command"
-    # recoding benchmarking command ang GPU command
-    jq_output=$(jq -n \
-      --arg command "$throughput_command" \
-      --arg gpu "$gpu_type" \
-      '{
-        throughput_command: $command,
-        gpu_type: $gpu
-      }')
-    echo "$jq_output" >"$RESULTS_FOLDER/$test_name.commands"
-
-    # run the benchmark
-    eval "$throughput_command"
-
-    kill_gpu_processes
-
-  done
-}
-
-run_serving_tests() {
-  # run serving tests using `benchmark_serving.py`
-  # $1: a json file specifying serving test cases
-
-  local serving_test_file
-  serving_test_file=$1
-
-  # Iterate over serving tests
-  jq -c '.[]' "$serving_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-    if [[ ! "$test_name" =~ ^serving_ ]]; then
-      echo "In serving-test.json, test_name must start with \"serving_\"."
-      exit 1
-    fi
-
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-
-    # get client and server arguments
-    server_params=$(echo "$params" | jq -r '.server_parameters')
-    client_params=$(echo "$params" | jq -r '.client_parameters')
-    server_args=$(json2args "$server_params")
-    client_args=$(json2args "$client_params")
-    qps_list=$(echo "$params" | jq -r '.qps_list')
-    qps_list=$(echo "$qps_list" | jq -r '.[] | @sh')
-    echo "Running over qps list $qps_list"
-
-    # check if there is enough GPU to run the test
-    tp=$(echo "$server_params" | jq -r '.tensor_parallel_size')
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $testname."
-      continue
-    fi
-
-    # check if server model and client model is aligned
-    server_model=$(echo "$server_params" | jq -r '.model')
-    client_model=$(echo "$client_params" | jq -r '.model')
-    if [[ $server_model != "$client_model" ]]; then
-      echo "Server model and client model must be the same. Skip testcase $testname."
-      continue
-    fi
-
-    server_command="python3 \
-      -m vllm.entrypoints.openai.api_server \
-      $server_args"
-
-    # run the server
-    echo "Running test case $test_name"
-    echo "Server command: $server_command"
-    eval "$server_command" &
-    server_pid=$!
-
-    # wait until the server is alive
-    wait_for_server
-    if [ $? -eq 0 ]; then
-      echo ""
-      echo "vllm server is up and running."
-    else
-      echo ""
-      echo "vllm failed to start within the timeout period."
-    fi
-
-    # iterate over different QPS
-    for qps in $qps_list; do
-      # remove the surrounding single quote from qps
-      if [[ "$qps" == *"inf"* ]]; then
-        echo "qps was $qps"
-        qps="inf"
-        echo "now qps is $qps"
-      fi
-
-      new_test_name=$test_name"_qps_"$qps
-
-      client_command="python3 benchmark_serving.py \
-        --save-result \
-        --result-dir $RESULTS_FOLDER \
-        --result-filename ${new_test_name}.json \
-        --request-rate $qps \
-        $client_args"
-
-      echo "Running test case $test_name with qps $qps"
-      echo "Client command: $client_command"
-
-      eval "$client_command"
-
-      # record the benchmarking commands
-      jq_output=$(jq -n \
-        --arg server "$server_command" \
-        --arg client "$client_command" \
-        --arg gpu "$gpu_type" \
-        '{
-          server_command: $server,
-          client_command: $client,
-          gpu_type: $gpu
-        }')
-      echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"
-
-    done
-
-    # clean up
-    kill -9 $server_pid
-    kill_gpu_processes
-  done
-}
-
-main() {
-  check_gpus
-  check_hf_token
-
-  # dependencies
-  (which wget && which curl) || (apt-get update && apt-get install -y wget curl)
-  (which jq) || (apt-get update && apt-get -y install jq)
-  (which lsof) || (apt-get update && apt-get install -y lsof)
-
-  # get the current IP address, required by benchmark_serving.py
-  export VLLM_HOST_IP=$(hostname -I | awk '{print $1}')
-  # turn of the reporting of the status of each request, to clean up the terminal output
-  export VLLM_LOG_LEVEL="WARNING"
-
-  # prepare for benchmarking
-  cd benchmarks || exit 1
-  ensure_sharegpt_downloaded
-  declare -g RESULTS_FOLDER=results/
-  mkdir -p $RESULTS_FOLDER
-  QUICK_BENCHMARK_ROOT=../.buildkite/nightly-benchmarks/
-
-  # benchmarking
-  run_serving_tests $QUICK_BENCHMARK_ROOT/tests/serving-tests.json
-  run_latency_tests $QUICK_BENCHMARK_ROOT/tests/latency-tests.json
-  run_throughput_tests $QUICK_BENCHMARK_ROOT/tests/throughput-tests.json
-
-  # postprocess benchmarking results
-  pip install tabulate pandas
-  python3 $QUICK_BENCHMARK_ROOT/scripts/convert-results-json-to-markdown.py
-
-  upload_to_buildkite
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/run-tgi-nightly.sh

@@ -1,216 +0,0 @@
-#!/bin/bash
-
-set -o pipefail
-
-check_gpus() {
-  # check the number of GPUs and GPU type.
-  declare -g gpu_count=$(nvidia-smi --list-gpus | wc -l)
-  if [[ $gpu_count -gt 0 ]]; then
-    echo "GPU found."
-  else
-    echo "Need at least 1 GPU to run benchmarking."
-    exit 1
-  fi
-  declare -g gpu_type=$(echo $(nvidia-smi --query-gpu=name --format=csv,noheader) | awk '{print $2}')
-  echo "GPU type is $gpu_type"
-}
-
-kill_gpu_processes() {
-  pkill text-generation || true
-  # waiting for GPU processes to be fully killed
-  sleep 10
-  # Print the GPU memory usage
-  # so that we know if all GPU processes are killed.
-  gpu_memory_usage=$(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits -i 0)
-  # The memory usage should be 0 MB.
-  echo "GPU 0 Memory Usage: $gpu_memory_usage MB"
-}
-
-json2args() {
-  # transforms the JSON string to command line args, and '_' is replaced to '-'
-  # example:
-  # input: { "model": "meta-llama/Llama-2-7b-chat-hf", "tensor_parallel_size": 1 }
-  # output: --model meta-llama/Llama-2-7b-chat-hf --tensor-parallel-size 1
-  local json_string=$1
-  local args=$(
-    echo "$json_string" | jq -r '
-      to_entries |
-      map("--" + (.key | gsub("_"; "-")) + " " + (.value | tostring)) |
-      join(" ")
-    '
-  )
-  echo "$args"
-}
-
-wait_for_server() {
-  timeout 1200 bash -c '
-    until curl -s localhost:8000/generate_stream > /dev/null; do
-      sleep 1
-    done' && return 0 || return 1
-}
-
-run_serving_tests() {
-  # run serving tests using `benchmark_serving.py`
-  # $1: a json file specifying serving test cases
-
-  local serving_test_file
-  serving_test_file=$1
-
-  # Iterate over serving tests
-  jq -c '.[]' "$serving_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-    
-
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-
-    # append tgi to the test name
-    test_name=tgi_$test_name
-
-    # get common parameters
-    common_params=$(echo "$params" | jq -r '.common_parameters')
-    model=$(echo "$common_params" | jq -r '.model')
-    tp=$(echo "$common_params" | jq -r '.tp')
-    dataset_name=$(echo "$common_params" | jq -r '.dataset_name')
-    dataset_path=$(echo "$common_params" | jq -r '.dataset_path')
-    port=$(echo "$common_params" | jq -r '.port')
-    num_prompts=$(echo "$common_params" | jq -r '.num_prompts')
-
-    # get client and server arguments
-    server_params=$(echo "$params" | jq -r '.tgi_server_parameters')
-    client_params=$(echo "$params" | jq -r '.tgi_client_parameters')
-    server_args=$(json2args "$server_params")
-    client_args=$(json2args "$client_params")
-    qps_list=$(echo "$params" | jq -r '.qps_list')
-    qps_list=$(echo "$qps_list" | jq -r '.[] | @sh')
-    echo "Running over qps list $qps_list"
-
-    # check if there is enough GPU to run the test
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required num-shard $tp but only $gpu_count GPU found. Skip testcase $test_name."
-      continue
-    fi
-
-    if echo "$common_params" | jq -e 'has("fp8")' > /dev/null; then
-      echo "Key 'fp8' exists in common params."
-      server_command="/tgi-entrypoint.sh \
-        --model-id $model \
-        --num-shard $tp \
-        --port $port \
-        --quantize fp8 \
-        $server_args"
-    else
-      echo "Key 'fp8' does not exist in common params."
-      server_command="/tgi-entrypoint.sh \
-        --model-id $model \
-        --num-shard $tp \
-        --port $port \
-        $server_args"
-    fi
-
-
-    
-
-    # run the server
-    echo "Running test case $test_name"
-    echo "Server command: $server_command"
-    eval "$server_command" &
-
-    # wait until the server is alive
-    wait_for_server
-    if [ $? -eq 0 ]; then
-      echo ""
-      echo "tgi server is up and running."
-    else
-      echo ""
-      echo "tgi failed to start within the timeout period."
-      break
-    fi
-
-    # iterate over different QPS
-    for qps in $qps_list; do
-      # remove the surrounding single quote from qps
-      if [[ "$qps" == *"inf"* ]]; then
-        echo "qps was $qps"
-        qps="inf"
-        echo "now qps is $qps"
-      fi
-
-      new_test_name=$test_name"_qps_"$qps
-
-      client_command="python3 benchmark_serving.py \
-        --backend tgi \
-        --model $model \
-        --dataset-name $dataset_name \
-        --dataset-path $dataset_path \
-        --num-prompts $num_prompts \
-        --port $port \
-        --save-result \
-        --result-dir $RESULTS_FOLDER \
-        --result-filename ${new_test_name}.json \
-        --request-rate $qps \
-        $client_args"
-
-      echo "Running test case $test_name with qps $qps"
-      echo "Client command: $client_command"
-
-      eval "$client_command"
-
-      # record the benchmarking commands
-      jq_output=$(jq -n \
-        --arg server "$server_command" \
-        --arg client "$client_command" \
-        --arg gpu "$gpu_type" \
-        --arg engine "tgi" \
-        '{
-          server_command: $server,
-          client_command: $client,
-          gpu_type: $gpu,
-          engine: $engine
-        }')
-      echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"
-
-    done
-
-    # clean up
-    kill_gpu_processes
-    rm -rf /root/.cache/huggingface/*
-  done
-}
-
-
-
-upload_to_buildkite() {
-  # upload the benchmarking results to buildkite
-
-  # if the agent binary is not found, skip uploading the results, exit 0
-  if [ ! -f /workspace/buildkite-agent ]; then
-    echo "buildkite-agent binary not found. Skip uploading the results."
-    return 0
-  fi
-  # /workspace/buildkite-agent annotate --style "success" --context "benchmark-results" --append < $RESULTS_FOLDER/${CURRENT_LLM_SERVING_ENGINE}_nightly_results.md
-  /workspace/buildkite-agent artifact upload "$RESULTS_FOLDER/*"
-}
-
-main() {
-
-  check_gpus
-  # enter vllm directory
-  cd $VLLM_SOURCE_CODE_LOC/benchmarks
-  declare -g RESULTS_FOLDER=results/
-  mkdir -p $RESULTS_FOLDER
-  BENCHMARK_ROOT=../.buildkite/nightly-benchmarks/
-
-  export CURRENT_LLM_SERVING_ENGINE=tgi
-  run_serving_tests $BENCHMARK_ROOT/tests/nightly-tests.json
-  python -m pip install tabulate pandas
-  python $BENCHMARK_ROOT/scripts/summary-nightly-results.py
-  upload_to_buildkite
-
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/run-trt-nightly.sh

@@ -1,214 +0,0 @@
-#!/bin/bash
-
-set -o pipefail
-
-check_gpus() {
-  # check the number of GPUs and GPU type.
-  declare -g gpu_count=$(nvidia-smi --list-gpus | wc -l)
-  if [[ $gpu_count -gt 0 ]]; then
-    echo "GPU found."
-  else
-    echo "Need at least 1 GPU to run benchmarking."
-    exit 1
-  fi
-  declare -g gpu_type=$(echo $(nvidia-smi --query-gpu=name --format=csv,noheader) | awk '{print $2}')
-  echo "GPU type is $gpu_type"
-}
-
-kill_gpu_processes() {
-  pkill tritonserver || true
-  # waiting for GPU processes to be fully killed
-  sleep 20
-  # Print the GPU memory usage
-  # so that we know if all GPU processes are killed.
-  gpu_memory_usage=$(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits -i 0)
-  # The memory usage should be 0 MB.
-  echo "GPU 0 Memory Usage: $gpu_memory_usage MB"
-}
-
-json2args() {
-  # transforms the JSON string to command line args, and '_' is replaced to '-'
-  # example:
-  # input: { "model": "meta-llama/Llama-2-7b-chat-hf", "tensor_parallel_size": 1 }
-  # output: --model meta-llama/Llama-2-7b-chat-hf --tensor-parallel-size 1
-  local json_string=$1
-  local args=$(
-    echo "$json_string" | jq -r '
-      to_entries |
-      map("--" + (.key | gsub("_"; "-")) + " " + (.value | tostring)) |
-      join(" ")
-    '
-  )
-  echo "$args"
-}
-
-wait_for_server() {
-  timeout 1200 bash -c '
-    until curl -s localhost:8000/generate_stream > /dev/null; do
-      sleep 1
-    done' && return 0 || return 1
-}
-
-run_serving_tests() {
-  # run serving tests using `benchmark_serving.py`
-  # $1: a json file specifying serving test cases
-
-  local serving_test_file
-  serving_test_file=$1
-
-  # Iterate over serving tests
-  jq -c '.[]' "$serving_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-    
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-
-    # append trt to the test name
-    test_name=trt_$test_name
-
-    # get common parameters
-    common_params=$(echo "$params" | jq -r '.common_parameters')
-    model=$(echo "$common_params" | jq -r '.model')
-    tp=$(echo "$common_params" | jq -r '.tp')
-    dataset_name=$(echo "$common_params" | jq -r '.dataset_name')
-    dataset_path=$(echo "$common_params" | jq -r '.dataset_path')
-    port=$(echo "$common_params" | jq -r '.port')
-    num_prompts=$(echo "$common_params" | jq -r '.num_prompts')
-
-    # get client and server arguments
-    server_params=$(echo "$params" | jq -r '.trt_server_parameters')
-    client_params=$(echo "$params" | jq -r '.trt_client_parameters')
-    client_args=$(json2args "$client_params")
-    qps_list=$(echo "$params" | jq -r '.qps_list')
-    qps_list=$(echo "$qps_list" | jq -r '.[] | @sh')
-    echo "Running over qps list $qps_list"
-
-    # check if there is enough GPU to run the test
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required model_tp_size $tp but only $gpu_count GPU found. Skip testcase $test_name."
-      continue
-    fi
-
-
-
-    cd $VLLM_SOURCE_CODE_LOC/benchmarks
-
-
-    echo "Running test case $test_name"
-    bash ../.buildkite/nightly-benchmarks/scripts/launch-trt-server.sh "$server_params" "$common_params"
-
-    # wait until the server is alive
-    wait_for_server
-    if [ $? -eq 0 ]; then
-      echo ""
-      echo "trt server is up and running."
-    else
-      echo ""
-      echo "trt failed to start within the timeout period."
-      break
-    fi
-
-    # prepare tokenizer
-    cd $VLLM_SOURCE_CODE_LOC/benchmarks
-    rm -rf /tokenizer_cache
-    mkdir /tokenizer_cache
-    python ../.buildkite/nightly-benchmarks/scripts/download-tokenizer.py \
-      --model "$model" \
-      --cachedir /tokenizer_cache
-    cd $VLLM_SOURCE_CODE_LOC/benchmarks
-    
-
-    # iterate over different QPS
-    for qps in $qps_list; do
-      # remove the surrounding single quote from qps
-      if [[ "$qps" == *"inf"* ]]; then
-        echo "qps was $qps"
-        qps="inf"
-        echo "now qps is $qps"
-      fi
-
-      new_test_name=$test_name"_qps_"$qps
-
-      client_command="python3 benchmark_serving.py \
-        --backend tensorrt-llm \
-        --tokenizer /tokenizer_cache \
-        --model $model \
-        --dataset-name $dataset_name \
-        --dataset-path $dataset_path \
-        --num-prompts $num_prompts \
-        --port $port \
-        --save-result \
-        --result-dir $RESULTS_FOLDER \
-        --result-filename ${new_test_name}.json \
-        --request-rate $qps \
-        $client_args"
-
-      echo "Running test case $test_name with qps $qps"
-      echo "Client command: $client_command"
-
-      eval "$client_command"
-
-      server_command=""
-      # record the benchmarking commands
-      jq_output=$(jq -n \
-        --arg server "$server_command" \
-        --arg client "$client_command" \
-        --arg gpu "$gpu_type" \
-        --arg engine "trt" \
-        '{
-          server_command: $server,
-          client_command: $client,
-          gpu_type: $gpu,
-          engine: $engine
-        }')
-      echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"
-
-    done
-
-    # clean up
-    kill_gpu_processes
-    rm -rf /root/.cache/huggingface/*
-  done
-}
-
-upload_to_buildkite() {
-  # upload the benchmarking results to buildkite
-
-  # if the agent binary is not found, skip uploading the results, exit 0
-  if [ ! -f /workspace/buildkite-agent ]; then
-    echo "buildkite-agent binary not found. Skip uploading the results."
-    return 0
-  fi
-  # /workspace/buildkite-agent annotate --style "success" --context "benchmark-results" --append < $RESULTS_FOLDER/${CURRENT_LLM_SERVING_ENGINE}_nightly_results.md
-  /workspace/buildkite-agent artifact upload "$RESULTS_FOLDER/*"
-}
-
-
-main() {
-
-  check_gpus
-
-
-  # enter vllm directory
-  cd $VLLM_SOURCE_CODE_LOC/benchmarks
-
-  declare -g RESULTS_FOLDER=results/
-  mkdir -p $RESULTS_FOLDER
-  BENCHMARK_ROOT=../.buildkite/nightly-benchmarks/
-
-  # update transformers package, to make sure mixtral tokenizer is available
-  python -m pip install transformers -U
-
-  export CURRENT_LLM_SERVING_ENGINE=trt
-  run_serving_tests $BENCHMARK_ROOT/tests/nightly-tests.json
-  python -m pip install tabulate pandas
-  python $BENCHMARK_ROOT/scripts/summary-nightly-results.py
-  upload_to_buildkite
-
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/run-vllm-nightly.sh

@@ -1,221 +0,0 @@
-#!/bin/bash
-
-set -o pipefail
-
-check_gpus() {
-  # check the number of GPUs and GPU type.
-  declare -g gpu_count=$(nvidia-smi --list-gpus | wc -l)
-  if [[ $gpu_count -gt 0 ]]; then
-    echo "GPU found."
-  else
-    echo "Need at least 1 GPU to run benchmarking."
-    exit 1
-  fi
-  declare -g gpu_type=$(echo $(nvidia-smi --query-gpu=name --format=csv,noheader) | awk '{print $2}')
-  echo "GPU type is $gpu_type"
-}
-
-kill_gpu_processes() {
-  # kill all processes on GPU.
-  pkill pt_main_thread
-  sleep 10
-
-  # remove vllm config file
-  rm -rf ~/.config/vllm
-
-  # Print the GPU memory usage
-  # so that we know if all GPU processes are killed.
-  gpu_memory_usage=$(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits -i 0)
-  # The memory usage should be 0 MB.
-  echo "GPU 0 Memory Usage: $gpu_memory_usage MB"
-}
-
-json2args() {
-  # transforms the JSON string to command line args, and '_' is replaced to '-'
-  # example:
-  # input: { "model": "meta-llama/Llama-2-7b-chat-hf", "tensor_parallel_size": 1 }
-  # output: --model meta-llama/Llama-2-7b-chat-hf --tensor-parallel-size 1
-  local json_string=$1
-  local args=$(
-    echo "$json_string" | jq -r '
-      to_entries |
-      map("--" + (.key | gsub("_"; "-")) + " " + (.value | tostring)) |
-      join(" ")
-    '
-  )
-  echo "$args"
-}
-
-wait_for_server() {
-  # wait for vllm server to start
-  # return 1 if vllm server crashes
-  timeout 1200 bash -c '
-    until curl -s localhost:8000/v1/completions > /dev/null; do
-      sleep 1
-    done' && return 0 || return 1
-}
-
-run_serving_tests() {
-  # run serving tests using `benchmark_serving.py`
-  # $1: a json file specifying serving test cases
-
-  local serving_test_file
-  serving_test_file=$1
-
-  # Iterate over serving tests
-  jq -c '.[]' "$serving_test_file" | while read -r params; do
-    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-
-    # if TEST_SELECTOR is set, only run the test cases that match the selector
-    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
-      echo "Skip test case $test_name."
-      continue
-    fi
-
-    # append vllm to the test name
-    test_name=vllm_$test_name
-
-
-    # get common parameters
-    common_params=$(echo "$params" | jq -r '.common_parameters')
-    model=$(echo "$common_params" | jq -r '.model')
-    tp=$(echo "$common_params" | jq -r '.tp')
-    dataset_name=$(echo "$common_params" | jq -r '.dataset_name')
-    dataset_path=$(echo "$common_params" | jq -r '.dataset_path')
-    port=$(echo "$common_params" | jq -r '.port')
-    num_prompts=$(echo "$common_params" | jq -r '.num_prompts')
-
-    # get client and server arguments
-    server_params=$(echo "$params" | jq -r '.vllm_server_parameters')
-    client_params=$(echo "$params" | jq -r '.vllm_client_parameters')
-    server_args=$(json2args "$server_params")
-    client_args=$(json2args "$client_params")
-    qps_list=$(echo "$params" | jq -r '.qps_list')
-    qps_list=$(echo "$qps_list" | jq -r '.[] | @sh')
-    echo "Running over qps list $qps_list"
-
-    # check if there is enough GPU to run the test
-    if [[ $gpu_count -lt $tp ]]; then
-      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
-      continue
-    fi
-
-    if echo "$common_params" | jq -e 'has("fp8")' > /dev/null; then
-      echo "Key 'fp8' exists in common params. Use neuralmagic fp8 model for convenience."
-      model=$(echo "$common_params" | jq -r '.neuralmagic_quantized_model')
-      server_command="python3 \
-        -m vllm.entrypoints.openai.api_server \
-        -tp $tp \
-        --model $model \
-        --port $port \
-        $server_args"
-    else
-      echo "Key 'fp8' does not exist in common params."
-      server_command="python3 \
-        -m vllm.entrypoints.openai.api_server \
-        -tp $tp \
-        --model $model \
-        --port $port \
-        $server_args"
-    fi
-
-    # run the server
-    echo "Running test case $test_name"
-    echo "Server command: $server_command"
-    eval "$server_command" &
-
-    # wait until the server is alive
-    wait_for_server
-    if [ $? -eq 0 ]; then
-      echo ""
-      echo "vllm server is up and running."
-    else
-      echo ""
-      echo "vllm failed to start within the timeout period."
-      break
-    fi
-
-    # iterate over different QPS
-    for qps in $qps_list; do
-      # remove the surrounding single quote from qps
-      if [[ "$qps" == *"inf"* ]]; then
-        echo "qps was $qps"
-        qps="inf"
-        echo "now qps is $qps"
-      fi
-
-      new_test_name=$test_name"_qps_"$qps
-
-      client_command="python3 benchmark_serving.py \
-        --backend vllm \
-        --model $model \
-        --dataset-name $dataset_name \
-        --dataset-path $dataset_path \
-        --num-prompts $num_prompts \
-        --port $port \
-        --save-result \
-        --result-dir $RESULTS_FOLDER \
-        --result-filename ${new_test_name}.json \
-        --request-rate $qps \
-        $client_args"
-
-      echo "Running test case $test_name with qps $qps"
-      echo "Client command: $client_command"
-
-      eval "$client_command"
-
-      # record the benchmarking commands
-      jq_output=$(jq -n \
-        --arg server "$server_command" \
-        --arg client "$client_command" \
-        --arg gpu "$gpu_type" \
-        --arg engine "vllm" \
-        '{
-          server_command: $server,
-          client_command: $client,
-          gpu_type: $gpu,
-          engine: $engine
-        }')
-      echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"
-
-    done
-
-    # clean up
-    kill_gpu_processes
-    rm -rf /root/.cache/huggingface/*
-  done
-}
-
-
-upload_to_buildkite() {
-  # upload the benchmarking results to buildkite
-
-  # if the agent binary is not found, skip uploading the results, exit 0
-  if [ ! -f /workspace/buildkite-agent ]; then
-    echo "buildkite-agent binary not found. Skip uploading the results."
-    return 0
-  fi
-  # /workspace/buildkite-agent annotate --style "success" --context "benchmark-results" --append < $RESULTS_FOLDER/${CURRENT_LLM_SERVING_ENGINE}_nightly_results.md
-  /workspace/buildkite-agent artifact upload "$RESULTS_FOLDER/*"
-}
-
-main() {
-
-  check_gpus
-  # enter vllm directory
-  cd $VLLM_SOURCE_CODE_LOC/benchmarks
-  declare -g RESULTS_FOLDER=results/
-  mkdir -p $RESULTS_FOLDER
-  BENCHMARK_ROOT=../.buildkite/nightly-benchmarks/
-
-  export CURRENT_LLM_SERVING_ENGINE=vllm
-  run_serving_tests $BENCHMARK_ROOT/tests/nightly-tests.json
-
-  python3 -m pip install tabulate pandas
-  python3 $BENCHMARK_ROOT/scripts/summary-nightly-results.py
-  upload_to_buildkite
-
-}
-
-main "$@"

.buildkite/nightly-benchmarks/scripts/summary-nightly-results.py

@@ -1,76 +0,0 @@
-import datetime
-import json
-import os
-from pathlib import Path
-
-import pandas as pd
-from tabulate import tabulate
-
-results_folder = Path("results/")
-
-# serving results and the keys that will be printed into markdown
-serving_results = []
-serving_column_mapping = {
-    "test_name": "Test name",
-    "gpu_type": "GPU",
-    "completed": "Successful req.",
-    "request_throughput": "Tput (req/s)",
-    "mean_ttft_ms": "Mean TTFT (ms)",
-    "std_ttft_ms": "Std TTFT (ms)",
-    "mean_itl_ms": "Mean ITL (ms)",
-    "std_itl_ms": "Std ITL (ms)",
-    "input_throughput": "Input Tput (tok/s)",
-    "output_throughput": "Output Tput (tok/s)",
-    "engine": "Engine",
-}
-
-if __name__ == "__main__":
-
-    # collect results
-    for test_file in results_folder.glob("*.json"):
-
-        with open(test_file, "r") as f:
-            raw_result = json.loads(f.read())
-
-        # attach the benchmarking command to raw_result
-        with open(test_file.with_suffix(".commands"), "r") as f:
-            command = json.loads(f.read())
-        raw_result.update(command)
-
-        # update the test name of this result
-        raw_result.update({"test_name": test_file.stem})
-
-        # add the result to raw_result
-        serving_results.append(raw_result)
-        continue
-
-    serving_results = pd.DataFrame.from_dict(serving_results)
-
-    if not serving_results.empty:
-        serving_results = serving_results[list(
-            serving_column_mapping.keys())].rename(
-                columns=serving_column_mapping)
-
-    serving_md_table_with_headers = tabulate(serving_results,
-                                             headers='keys',
-                                             tablefmt='pipe',
-                                             showindex=False)
-    # remove the first line of header
-    serving_md_table_lines = serving_md_table_with_headers.split('\n')
-    serving_md_table_without_header = '\n'.join(serving_md_table_lines[2:])
-
-    prefix = datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
-    prefix = prefix + "_" + os.environ.get("CURRENT_LLM_SERVING_ENGINE")
-
-    # document benchmarking results in markdown
-    with open(results_folder / f"{prefix}_nightly_results.md", "w") as f:
-        # document results with header.
-        # for those who wants to reproduce our benchmark.
-        f.write(serving_md_table_with_headers)
-        f.write('\n')
-
-    # document benchmarking results in json
-    with open(results_folder / f"{prefix}_nightly_results.json", "w") as f:
-
-        results = serving_results.to_dict(orient='records')
-        f.write(json.dumps(results))

.buildkite/nightly-benchmarks/scripts/wait-for-image.sh

@@ -1,19 +0,0 @@
-#!/bin/sh
-TOKEN=$(curl -s -L "https://public.ecr.aws/token?service=public.ecr.aws&scope=repository:q9t5s3a7/vllm-ci-test-repo:pull" | jq -r .token)
-URL="https://public.ecr.aws/v2/q9t5s3a7/vllm-ci-test-repo/manifests/$BUILDKITE_COMMIT"
-
-TIMEOUT_SECONDS=10
-
-retries=0
-while [ $retries -lt 1000 ]; do
-    if [ $(curl -s --max-time $TIMEOUT_SECONDS -L -H "Authorization: Bearer $TOKEN" -o /dev/null -w "%{http_code}" $URL) -eq 200 ]; then
-        exit 0
-    fi
-
-    echo "Waiting for image to be available..."
-
-    retries=$((retries + 1))
-    sleep 5
-done
-
-exit 1

.buildkite/nightly-benchmarks/tests/latency-tests.json

@@ -1,32 +0,0 @@
-[
-    {
-        "test_name": "latency_llama8B_tp1",
-        "parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-            "load_format": "dummy",
-            "num_iters_warmup": 5,
-            "num_iters": 15
-        }
-    },
-    {
-        "test_name": "latency_llama70B_tp4",
-        "parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-            "tensor_parallel_size": 4,
-            "load_format": "dummy",
-            "num-iters-warmup": 5,
-            "num-iters": 15
-        }
-    },
-    {
-        "test_name": "latency_mixtral8x7B_tp2",
-        "parameters": {
-            "model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            "tensor_parallel_size": 2,
-            "load_format": "dummy",
-            "num-iters-warmup": 5,
-            "num-iters": 15
-        }
-    }
-]

.buildkite/nightly-benchmarks/tests/nightly-tests.json

@@ -1,116 +0,0 @@
-[
-    {
-        "test_name": "llama8B_tp1",
-        "qps_list": [4],
-        "common_parameters": {
-            "model": "meta-llama/Meta-Llama-3-8B",
-            "tp": 1,
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 500,
-            "port": 8000
-        },
-        "lmdeploy_server_parameters": {
-        },
-        "lmdeploy_client_parameters": {
-        },
-        "tgi_server_parameters": {
-        },
-        "tgi_client_parameters": {
-            "endpoint": "/generate_stream"
-        },
-        "trt_server_parameters": {
-            "model_type": "llama",
-            "model_dtype": "float16",
-            "max_batch_size": 256,
-            "max_input_len": 4096,
-            "max_output_len": 4096,
-            "trt_llm_version": "r24.04"
-        },
-        "trt_client_parameters": {
-            "endpoint": "/v2/models/ensemble/generate_stream"
-        },
-        "vllm_server_parameters": {
-            "disable_log_stats": "",
-            "disable_log_requests": ""
-        },
-        "vllm_client_parameters": {
-        }
-    },
-    {
-        "test_name": "llama70B_tp4",
-        "qps_list": [2],
-        "common_parameters": {
-            "model": "meta-llama/Meta-Llama-3-70B-Instruct",
-            "tp": 4,
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 500,
-            "port": 8000
-        },
-        "lmdeploy_server_parameters": {
-        },
-        "lmdeploy_client_parameters": {
-        },
-        "tgi_server_parameters": {
-        },
-        "tgi_client_parameters": {
-            "endpoint": "/generate_stream"
-        },
-        "trt_server_parameters": {
-            "model_type": "llama",
-            "model_dtype": "float16",
-            "max_batch_size": 256,
-            "max_input_len": 4096,
-            "max_output_len": 4096,
-            "trt_llm_version": "r24.04"
-        },
-        "trt_client_parameters": {
-            "endpoint": "/v2/models/ensemble/generate_stream"
-        },
-        "vllm_server_parameters": {
-            "disable_log_stats": "",
-            "disable_log_requests": ""
-        },
-        "vllm_client_parameters": {
-        }
-    },
-    {
-        "test_name": "mixtral8x7B_tp2",
-        "qps_list": [2],
-        "common_parameters": {
-            "model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            "tp": 2,
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 500,
-            "port": 8000
-        },
-        "lmdeploy_server_parameters": {
-        },
-        "lmdeploy_client_parameters": {
-        },
-        "tgi_server_parameters": {
-        },
-        "tgi_client_parameters": {
-            "endpoint": "/generate_stream"
-        },
-        "trt_server_parameters": {
-            "model_type": "llama",
-            "model_dtype": "float16",
-            "max_batch_size": 256,
-            "max_input_len": 4096,
-            "max_output_len": 4096,
-            "trt_llm_version": "r24.04"
-        },
-        "trt_client_parameters": {
-            "endpoint": "/v2/models/ensemble/generate_stream"
-        },
-        "vllm_server_parameters": {
-            "disable_log_stats": "",
-            "disable_log_requests": ""
-        },
-        "vllm_client_parameters": {
-        }
-    }
-]

.buildkite/nightly-benchmarks/tests/serving-tests.json

@@ -1,80 +0,0 @@
-[
-    {
-        "test_name": "serving_llama8B_tp1_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "server_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-            "swap_space": 16,
-            "disable_log_stats": "",
-            "disable_log_requests": "",
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama70B_tp4_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "server_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-            "tensor_parallel_size": 4,
-            "swap_space": 16,
-            "disable_log_stats": "",
-            "disable_log_requests": "",
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_mixtral8x7B_tp2_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "server_parameters": {
-            "model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            "tensor_parallel_size": 2,
-            "swap_space": 16,
-            "disable_log_stats": "",
-            "disable_log_requests": "",
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama70B_tp4_sharegpt_specdecode",
-        "qps_list": [2],
-        "server_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-            "disable_log_requests": "", 
-            "tensor_parallel_size": 4,
-            "swap_space": 16, 
-            "speculative_model": "turboderp/Qwama-0.5B-Instruct",
-            "num_speculative_tokens": 4,
-            "speculative_draft_tensor_parallel_size": 1,
-            "use_v2_block_manager": ""
-        },
-        "client_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200 
-        }
-    }
-]

.buildkite/nightly-benchmarks/tests/throughput-tests.json

@@ -1,35 +0,0 @@
-[
-    {
-        "test_name": "throughput_llama8B_tp1",
-        "parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-            "load_format": "dummy",
-            "dataset": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200,
-            "backend": "vllm"
-        }
-    },
-    {
-        "test_name": "throughput_llama70B_tp4",
-        "parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-            "tensor_parallel_size": 4,
-            "load_format": "dummy",
-            "dataset": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200,
-            "backend": "vllm"
-        }
-    },
-    {
-        "test_name": "throughput_mixtral8x7B_tp2",
-        "parameters": {
-            "model": "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            "tensor_parallel_size": 2,
-            "load_format": "dummy",
-            "dataset": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200,
-            "backend": "vllm"
-        }
-    }
-]

.buildkite/release-pipeline.yaml

@@ -1,32 +0,0 @@
-steps:
-  - label: "Build wheel - CUDA 12.1"
-    agents:
-      queue: cpu_queue
-    commands:
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg buildkite_commit=$BUILDKITE_COMMIT --build-arg USE_SCCACHE=1 --build-arg CUDA_VERSION=12.1.0 --tag vllm-ci:build-image --target build --progress plain ."
-      - "mkdir artifacts"
-      - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      # rename the files to change linux -> manylinux1
-      - "for f in artifacts/dist/*.whl; do mv -- \"$$f\" \"$${f/linux/manylinux1}\"; done"
-      - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/$BUILDKITE_COMMIT/"
-      - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/nightly/"
-    env:
-      DOCKER_BUILDKIT: "1"
-
-  - block: "Build CUDA 11.8 wheel"
-    key: block-build-cu118-wheel
-  
-  - label: "Build wheel - CUDA 11.8"
-    depends_on: block-build-cu118-wheel
-    agents:
-      queue: cpu_queue
-    commands:
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg buildkite_commit=$BUILDKITE_COMMIT --build-arg USE_SCCACHE=1 --build-arg CUDA_VERSION=11.8.0 --tag vllm-ci:build-image --target build --progress plain ."
-      - "mkdir artifacts"
-      - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      # rename the files to change linux -> manylinux1
-      - "for f in artifacts/dist/*.whl; do mv -- \"$$f\" \"$${f/linux/manylinux1}\"; done"
-      - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/$BUILDKITE_COMMIT/"
-      - "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/nightly/"
-    env:
-      DOCKER_BUILDKIT: "1"

.buildkite/run-amd-test.sh

@@ -1,154 +1,38 @@
-# This script runs test inside the corresponding ROCm docker container.
-set -o pipefail
+# This script build the ROCm docker image and run the API server inside the container.
+# It serves a sanity check for compilation and basic model usage.
+set -ex
 
 # Print ROCm version
-echo "--- Confirming Clean Initial State"
-while true; do
-        sleep 3
-        if grep -q clean /opt/amdgpu/etc/gpu_state; then
-                echo "GPUs state is \"clean\""
-                break
-        fi
-done
-
-echo "--- ROCm info"
 rocminfo
 
-# cleanup older docker images
-cleanup_docker() {
-  # Get Docker's root directory
-  docker_root=$(docker info -f '{{.DockerRootDir}}')
-  if [ -z "$docker_root" ]; then
-    echo "Failed to determine Docker root directory."
-    exit 1
-  fi
-  echo "Docker root directory: $docker_root"
-  # Check disk usage of the filesystem where Docker's root directory is located
-  disk_usage=$(df "$docker_root" | tail -1 | awk '{print $5}' | sed 's/%//')
-  # Define the threshold
-  threshold=70
-  if [ "$disk_usage" -gt "$threshold" ]; then
-    echo "Disk usage is above $threshold%. Cleaning up Docker images and volumes..."
-    # Remove dangling images (those that are not tagged and not used by any container)
-    docker image prune -f
-    # Remove unused volumes
-    docker volume prune -f
-    echo "Docker images and volumes cleanup completed."
-  else
-    echo "Disk usage is below $threshold%. No cleanup needed."
-  fi
-}
+# Try building the docker image
+docker build -t rocm -f Dockerfile.rocm .
 
-# Call the cleanup docker function
-cleanup_docker
-
-echo "--- Resetting GPUs"
-
-echo "reset" > /opt/amdgpu/etc/gpu_state
-
-while true; do
-        sleep 3
-        if grep -q clean /opt/amdgpu/etc/gpu_state; then
-                echo "GPUs state is \"clean\""
-                break
-        fi
-done
-
-echo "--- Pulling container" 
-image_name="rocm/vllm-ci:${BUILDKITE_COMMIT}"
-container_name="rocm_${BUILDKITE_COMMIT}_$(tr -dc A-Za-z0-9 < /dev/urandom | head -c 10; echo)"
-docker pull ${image_name}
-
-remove_docker_container() {
-   docker rm -f ${container_name} || docker image rm -f ${image_name} || true
-}
+# Setup cleanup
+remove_docker_container() { docker rm -f rocm || true; }
 trap remove_docker_container EXIT
+remove_docker_container
 
-echo "--- Running container"
+# Run the image
+docker run --device /dev/kfd --device /dev/dri --network host --name rocm rocm python3 -m vllm.entrypoints.api_server &
 
-HF_CACHE="$(realpath ~)/huggingface"
-mkdir -p ${HF_CACHE}
-HF_MOUNT="/root/.cache/huggingface"
+# Wait for the server to start
+wait_for_server_to_start() {
+    timeout=300
+    counter=0
 
-commands=$@
-echo "Commands:$commands"
-#ignore certain kernels tests
-if [[ $commands == *" kernels "* ]]; then
-  commands="${commands} \
-  --ignore=kernels/test_attention.py \
-  --ignore=kernels/test_attention_selector.py \
-  --ignore=kernels/test_blocksparse_attention.py \
-  --ignore=kernels/test_causal_conv1d.py \
-  --ignore=kernels/test_cutlass.py \
-  --ignore=kernels/test_encoder_decoder_attn.py \
-  --ignore=kernels/test_flash_attn.py \
-  --ignore=kernels/test_flashinfer.py \
-  --ignore=kernels/test_gguf.py \
-  --ignore=kernels/test_int8_quant.py \
-  --ignore=kernels/test_machete_gemm.py \
-  --ignore=kernels/test_mamba_ssm.py \
-  --ignore=kernels/test_marlin_gemm.py \
-  --ignore=kernels/test_moe.py \
-  --ignore=kernels/test_prefix_prefill.py \
-  --ignore=kernels/test_rand.py \
-  --ignore=kernels/test_sampler.py"
-fi
+    while [ "$(curl -s -o /dev/null -w ''%{http_code}'' localhost:8000/health)" != "200" ]; do
+        sleep 1
+        counter=$((counter + 1))
+        if [ $counter -ge $timeout ]; then
+            echo "Timeout after $timeout seconds"
+            break
+        fi
+    done
+}
+wait_for_server_to_start
 
-#ignore certain Entrypoints tests
-if [[ $commands == *" entrypoints/openai "* ]]; then
-  commands=${commands//" entrypoints/openai "/" entrypoints/openai \
-  --ignore=entrypoints/openai/test_accuracy.py \
-  --ignore=entrypoints/openai/test_audio.py \
-  --ignore=entrypoints/openai/test_encoder_decoder.py \
-  --ignore=entrypoints/openai/test_embedding.py \
-  --ignore=entrypoints/openai/test_oot_registration.py "}
-fi
-
-PARALLEL_JOB_COUNT=8
-# check if the command contains shard flag, we will run all shards in parallel because the host have 8 GPUs. 
-if [[ $commands == *"--shard-id="* ]]; then
-  for GPU in $(seq 0 $(($PARALLEL_JOB_COUNT-1))); do
-    #replace shard arguments
-    commands=${commands//"--shard-id= "/"--shard-id=${GPU} "}
-    commands=${commands//"--num-shards= "/"--num-shards=${PARALLEL_JOB_COUNT} "}
-    echo "Shard ${GPU} commands:$commands"
-    docker run \
-        --device /dev/kfd --device /dev/dri \
-        --network host \
-        --shm-size=16gb \
-        --rm \
-        -e HIP_VISIBLE_DEVICES=${GPU} \
-        -e HF_TOKEN \
-        -v ${HF_CACHE}:${HF_MOUNT} \
-        -e HF_HOME=${HF_MOUNT} \
-        --name ${container_name}_${GPU}  \
-        ${image_name} \
-        /bin/bash -c "${commands}" \
-        |& while read -r line; do echo ">>Shard $GPU: $line"; done &
-    PIDS+=($!)
-  done
-  #wait for all processes to finish and collect exit codes
-  for pid in ${PIDS[@]}; do
-    wait ${pid}
-    STATUS+=($?)
-  done
-  for st in ${STATUS[@]}; do
-    if [[ ${st} -ne 0 ]]; then
-      echo "One of the processes failed with $st"
-      exit ${st}
-    fi
-  done
-else
-  docker run \
-          --device /dev/kfd --device /dev/dri \
-          --network host \
-          --shm-size=16gb \
-          --rm \
-          -e HIP_VISIBLE_DEVICES=0 \
-          -e HF_TOKEN \
-          -v ${HF_CACHE}:${HF_MOUNT} \
-          -e HF_HOME=${HF_MOUNT} \
-          --name ${container_name} \
-          ${image_name} \
-          /bin/bash -c "${commands}"
-fi
+# Test a simple prompt
+curl -X POST -H "Content-Type: application/json" \
+    localhost:8000/generate \
+    -d '{"prompt": "San Francisco is a"}'

.buildkite/run-benchmarks.sh

@@ -9,10 +9,10 @@ cd "$(dirname "${BASH_SOURCE[0]}")/.."
 (which wget && which curl) || (apt-get update && apt-get install -y wget curl)
 
 # run python-based benchmarks and upload the result to buildkite
-python3 benchmarks/benchmark_latency.py --output-json latency_results.json 2>&1 | tee benchmark_latency.txt
+python3 benchmarks/benchmark_latency.py 2>&1 | tee benchmark_latency.txt
 bench_latency_exit_code=$?
 
-python3 benchmarks/benchmark_throughput.py --input-len 256 --output-len 256 --output-json throughput_results.json 2>&1 | tee benchmark_throughput.txt
+python3 benchmarks/benchmark_throughput.py --input-len 256 --output-len 256 2>&1 | tee benchmark_throughput.txt
 bench_throughput_exit_code=$?
 
 # run server-based benchmarks and upload the result to buildkite
@@ -50,16 +50,11 @@ echo "### Serving Benchmarks" >> benchmark_results.md
 sed -n '1p' benchmark_serving.txt >> benchmark_results.md # first line
 echo "" >> benchmark_results.md
 echo '```' >> benchmark_results.md
-tail -n 24 benchmark_serving.txt >> benchmark_results.md # last 24 lines
+tail -n 20 benchmark_serving.txt >> benchmark_results.md # last 20 lines
 echo '```' >> benchmark_results.md
 
-# if the agent binary is not found, skip uploading the results, exit 0
-if [ ! -f /usr/bin/buildkite-agent ]; then
-    exit 0
-fi
-
 # upload the results to buildkite
-buildkite-agent annotate --style "info" --context "benchmark-results" < benchmark_results.md
+/workspace/buildkite-agent annotate --style "info" --context "benchmark-results" < benchmark_results.md
 
 # exit with the exit code of the benchmarks
 if [ $bench_latency_exit_code -ne 0 ]; then
@@ -74,5 +69,4 @@ if [ $bench_serving_exit_code -ne 0 ]; then
     exit $bench_serving_exit_code
 fi
 
-rm ShareGPT_V3_unfiltered_cleaned_split.json
-buildkite-agent artifact upload "*.json"
+/workspace/buildkite-agent artifact upload openai-*.json

.buildkite/run-cpu-test-ppc64le.sh

@@ -1,33 +0,0 @@
-# This script build the CPU docker image and run the offline inference inside the container.
-# It serves a sanity check for compilation and basic model usage.
-set -ex
-
-# Try building the docker image
-docker build -t cpu-test -f Dockerfile.ppc64le .
-
-# Setup cleanup
-remove_docker_container() { docker rm -f cpu-test || true; }
-trap remove_docker_container EXIT
-remove_docker_container
-
-# Run the image, setting --shm-size=4g for tensor parallel.
-source /etc/environment
-#docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test cpu-test
-docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN=$HF_TOKEN --name cpu-test cpu-test
-
-# Run basic model test
-docker exec cpu-test bash -c "
-  pip install pytest matplotlib einops transformers_stream_generator
-  pytest -v -s tests/models -m \"not vlm\" --ignore=tests/models/test_embedding.py --ignore=tests/models/test_oot_registration.py --ignore=tests/models/test_registry.py --ignore=tests/models/test_jamba.py --ignore=tests/models/test_danube3_4b.py" # Mamba and Danube3-4B on CPU is not supported
-
-# online inference
-docker exec cpu-test bash -c "
-  python3 -m vllm.entrypoints.openai.api_server --model facebook/opt-125m & 
-  timeout 600 bash -c 'until curl localhost:8000/v1/models; do sleep 1; done' || exit 1
-  python3 benchmarks/benchmark_serving.py \
-    --backend vllm \
-    --dataset-name random \
-    --model facebook/opt-125m \
-    --num-prompts 20 \
-    --endpoint /v1/completions \
-    --tokenizer facebook/opt-125m"

.buildkite/run-cpu-test.sh

@@ -3,47 +3,12 @@
 set -ex
 
 # Try building the docker image
-numactl -C 48-95 -N 1 docker build -t cpu-test -f Dockerfile.cpu .
-numactl -C 48-95 -N 1 docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" -t cpu-test-avx2 -f Dockerfile.cpu .
+docker build -t cpu-test -f Dockerfile.cpu .
 
 # Setup cleanup
-remove_docker_container() { docker rm -f cpu-test cpu-test-avx2 || true; }
+remove_docker_container() { docker rm -f cpu-test || true; }
 trap remove_docker_container EXIT
 remove_docker_container
 
-# Run the image, setting --shm-size=4g for tensor parallel.
-docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --cpuset-cpus=48-95 \
- --cpuset-mems=1 --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test cpu-test
-docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --cpuset-cpus=48-95 \
- --cpuset-mems=1 --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test-avx2 cpu-test-avx2
-
-# offline inference
-docker exec cpu-test-avx2 bash -c "python3 examples/offline_inference.py"
-
-# Run basic model test
-docker exec cpu-test bash -c "
-  pip install pytest matplotlib einops transformers_stream_generator datamodel_code_generator
-  pytest -v -s tests/models/decoder_only/language \
-    --ignore=tests/models/test_fp8.py \
-    --ignore=tests/models/decoder_only/language/test_jamba.py \
-    --ignore=tests/models/decoder_only/language/test_danube3_4b.py" # Mamba and Danube3-4B on CPU is not supported
-
-# Run compressed-tensor test
-docker exec cpu-test bash -c "
-  pytest -s -v \
-  tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_static_setup \
-  tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_dynanmic_per_token"
-
-# online inference
-docker exec cpu-test bash -c "
-  export VLLM_CPU_KVCACHE_SPACE=10 
-  export VLLM_CPU_OMP_THREADS_BIND=48-92 
-  python3 -m vllm.entrypoints.openai.api_server --model facebook/opt-125m & 
-  timeout 600 bash -c 'until curl localhost:8000/v1/models; do sleep 1; done' || exit 1
-  python3 benchmarks/benchmark_serving.py \
-    --backend vllm \
-    --dataset-name random \
-    --model facebook/opt-125m \
-    --num-prompts 20 \
-    --endpoint /v1/completions \
-    --tokenizer facebook/opt-125m"
+# Run the image and launch offline inference
+docker run --network host --env VLLM_CPU_KVCACHE_SPACE=1 --name cpu-test cpu-test python3 examples/offline_inference.py

.buildkite/run-multi-node-test.sh

@@ -1,105 +0,0 @@
-#!/bin/bash
-
-set -euox pipefail
-
-if [[ $# -lt 4 ]]; then
-    echo "Usage: .buildkite/run-multi-node-test.sh WORKING_DIR NUM_NODES NUM_GPUS DOCKER_IMAGE COMMAND1 COMMAND2 ... COMMANDN"
-    exit 1
-fi
-
-WORKING_DIR=$1
-NUM_NODES=$2
-NUM_GPUS=$3
-DOCKER_IMAGE=$4
-
-shift 4
-COMMANDS=("$@")
-if [ ${#COMMANDS[@]} -ne $NUM_NODES ]; then
-    echo "The number of commands must be equal to the number of nodes."
-    echo "Number of nodes: $NUM_NODES"
-    echo "Number of commands: ${#COMMANDS[@]}"
-    exit 1
-fi
-
-echo "List of commands"
-for command in "${COMMANDS[@]}"; do
-    echo $command
-done
-
-start_network() {
-    docker network create --subnet=192.168.10.0/24 docker-net
-}
-
-start_nodes() {
-    for node in $(seq 0 $(($NUM_NODES-1))); do
-        GPU_DEVICES='"device='
-        for node_gpu in $(seq 0 $(($NUM_GPUS - 1))); do
-            DEVICE_NUM=$(($node * $NUM_GPUS + $node_gpu))
-            GPU_DEVICES+=$(($DEVICE_NUM))
-            if [ $node_gpu -lt $(($NUM_GPUS - 1)) ]; then
-                GPU_DEVICES+=','
-            fi
-        done
-        GPU_DEVICES+='"'
-
-        # start the container in detached mode
-        # things to note:
-        # 1. --shm-size=10.24gb is required. don't use --ipc=host
-        # 2. pass HF_TOKEN to the container
-        # 3. map the huggingface cache directory to the container
-        # 3. assign ip addresses to the containers (head node: 192.168.10.10, worker nodes:
-        #    starting from 192.168.10.11)
-        docker run -d --gpus "$GPU_DEVICES" --shm-size=10.24gb -e HF_TOKEN -v ~/.cache/huggingface:/root/.cache/huggingface --name node$node --network docker-net --ip 192.168.10.$((10 + $node)) --rm $DOCKER_IMAGE /bin/bash -c "tail -f /dev/null"
-
-        # organize containers into a ray cluster
-        if [ $node -eq 0 ]; then
-            # start the ray head node
-            docker exec -d node$node /bin/bash -c "ray start --head --port=6379 --block"
-            # wait for the head node to be ready
-            sleep 10
-        else
-            # start the ray worker nodes, and connect them to the head node
-            docker exec -d node$node /bin/bash -c "ray start --address=192.168.10.10:6379 --block"
-        fi
-    done
-
-    # wait for the cluster to be ready
-    sleep 10
-
-    # print the cluster status
-    docker exec node0 /bin/bash -c "ray status"
-}
-
-run_nodes() {
-    # important: iterate in reverse order to start the head node last
-    # we start the worker nodes first, in detached mode, and then start the head node
-    # in the foreground, so that the output of the head node is visible in the buildkite logs
-    for node in $(seq $(($NUM_NODES - 1)) -1 0); do
-        GPU_DEVICES='"device='
-        for node_gpu in $(seq 0 $(($NUM_GPUS - 1))); do
-            DEVICE_NUM=$(($node * $NUM_GPUS + $node_gpu))
-            GPU_DEVICES+=$(($DEVICE_NUM))
-            if [ $node_gpu -lt $(($NUM_GPUS - 1)) ]; then
-                GPU_DEVICES+=','
-            fi
-        done
-        GPU_DEVICES+='"'
-        echo "Running node$node with GPU devices: $GPU_DEVICES"
-        if [ $node -ne 0 ]; then
-            docker exec -d node$node /bin/bash -c "cd $WORKING_DIR ; ${COMMANDS[$node]}"
-        else
-            docker exec node$node /bin/bash -c "cd $WORKING_DIR ; ${COMMANDS[$node]}"
-        fi
-    done
-}
-cleanup() {
-    for node in $(seq 0 $(($NUM_NODES-1))); do
-        docker stop node$node
-    done
-    docker network rm docker-net
-}
-trap cleanup EXIT
-start_network
-start_nodes
-run_nodes
-

.buildkite/run-neuron-test.sh

@@ -1,51 +0,0 @@
-# This script build the Neuron docker image and run the API server inside the container.
-# It serves a sanity check for compilation and basic model usage.
-set -e
-
-# Try building the docker image
-aws ecr get-login-password --region us-west-2 | docker login --username AWS --password-stdin 763104351884.dkr.ecr.us-west-2.amazonaws.com
-
-# prune old image and containers to save disk space, and only once a day
-# by using a timestamp file in tmp.
-if [ -f /tmp/neuron-docker-build-timestamp ]; then
-    last_build=$(cat /tmp/neuron-docker-build-timestamp)
-    current_time=$(date +%s)
-    if [ $((current_time - last_build)) -gt 86400 ]; then
-        docker system prune -f
-        echo $current_time > /tmp/neuron-docker-build-timestamp
-    fi
-else
-    echo $(date +%s) > /tmp/neuron-docker-build-timestamp
-fi
-
-docker build -t neuron -f Dockerfile.neuron .
-
-# Setup cleanup
-remove_docker_container() { docker rm -f neuron || true; }
-trap remove_docker_container EXIT
-remove_docker_container
-
-# Run the image
-docker run --device=/dev/neuron0 --device=/dev/neuron1 --network host --name neuron neuron python3 -m vllm.entrypoints.api_server \
-       --model TinyLlama/TinyLlama-1.1B-Chat-v1.0 --max-num-seqs 8 --max-model-len 128 --block-size 128 --device neuron --tensor-parallel-size 2 &
-
-# Wait for the server to start
-wait_for_server_to_start() {
-    timeout=300
-    counter=0
-
-    while [ "$(curl -s -o /dev/null -w ''%{http_code}'' localhost:8000/health)" != "200" ]; do
-        sleep 1
-        counter=$((counter + 1))
-        if [ $counter -ge $timeout ]; then
-            echo "Timeout after $timeout seconds"
-            break
-        fi
-    done
-}
-wait_for_server_to_start
-
-# Test a simple prompt
-curl -X POST -H "Content-Type: application/json" \
-    localhost:8000/generate \
-    -d '{"prompt": "San Francisco is a"}'

.buildkite/run-openvino-test.sh

@@ -1,14 +0,0 @@
-# This script build the OpenVINO docker image and run the offline inference inside the container.
-# It serves a sanity check for compilation and basic model usage.
-set -ex
-
-# Try building the docker image
-docker build -t openvino-test -f Dockerfile.openvino .
-
-# Setup cleanup
-remove_docker_container() { docker rm -f openvino-test || true; }
-trap remove_docker_container EXIT
-remove_docker_container
-
-# Run the image and launch offline inference
-docker run --network host --env VLLM_OPENVINO_KVCACHE_SPACE=1 --name openvino-test openvino-test python3 /workspace/vllm/examples/offline_inference.py

.buildkite/run-tpu-test.sh

@@ -1,15 +0,0 @@
-set -e
-
-# Build the docker image.
-docker build -f Dockerfile.tpu -t vllm-tpu .
-
-# Set up cleanup.
-remove_docker_container() { docker rm -f tpu-test || true; }
-trap remove_docker_container EXIT
-# Remove the container that might not be cleaned up in the previous run.
-remove_docker_container
-
-# For HF_TOKEN.
-source /etc/environment
-# Run a simple end-to-end example.
-docker run --privileged --net host --shm-size=16G -it -e HF_TOKEN=$HF_TOKEN --name tpu-test vllm-tpu /bin/bash -c "python3 -m pip install git+https://github.com/thuml/depyf.git && python3 -m pip install pytest  && pytest -v -s /workspace/vllm/tests/tpu/test_custom_dispatcher.py && python3 /workspace/vllm/tests/tpu/test_compilation.py && python3 /workspace/vllm/examples/offline_inference_tpu.py"

.buildkite/run-xpu-test.sh

@@ -1,14 +0,0 @@
-# This script build the CPU docker image and run the offline inference inside the container.
-# It serves a sanity check for compilation and basic model usage.
-set -ex
-
-# Try building the docker image
-docker build -t xpu-test -f Dockerfile.xpu .
-
-# Setup cleanup
-remove_docker_container() { docker rm -f xpu-test || true; }
-trap remove_docker_container EXIT
-remove_docker_container
-
-# Run the image and launch offline inference
-docker run --network host --name xpu-test --device /dev/dri -v /dev/dri/by-path:/dev/dri/by-path xpu-test python3 examples/offline_inference.py

.buildkite/test-pipeline.yaml

@@ -1,475 +1,111 @@
 # In this file, you can add more tests to run either by adding a new step or
 # adding a new command to an existing step. See different options here for examples.
-
-# This script will be feed into Jinja template in `test-template-aws.j2` at
-# https://github.com/vllm-project/buildkite-ci/blob/main/scripts/test-template-aws.j2 
-# to generate the final pipeline yaml file.
-
-# Documentation
-# label(str): the name of the test. emoji allowed.
-# fast_check(bool): whether to run this on each commit on fastcheck pipeline.
-# fast_check_only(bool): run this test on fastcheck pipeline only
-# command(str): the single command to run for tests. incompatible with commands.
-# commands(list): the list of commands to run for test. incompatbile with command.
-# mirror_hardwares(list): the list of hardwares to run the test on as well. currently only supports [amd]
-# gpu(str): override the GPU selection for the test. default is on L4 GPUs. currently only supports a100
-# num_gpus(int): override the number of GPUs for the test. default to 1 GPU. currently support 2,4.
-# num_nodes(int): whether to simulate multi-node setup by launch multiple containers on one host, 
-#     in this case, commands must be specified. the first command runs on first host, the second
-#     command runs on the second host.
-# working_dir(str): specify the place where command should execute, default to /vllm-workspace/tests
-# source_file_dependencies(list): the list of prefix to opt-in the test for, if empty, the test will always run.
-
-# When adding a test
-# - If the test belong to an existing group, add it there
-# - If the test is short, add to any existing step
-# - If the test takes more than 10min, then it is okay to create a new step. 
-#   Note that all steps execute in parallel. 
+# This script will be feed into Jinja template in `test-template.j2` to generate
+# the final pipeline yaml file.
 
 steps:
-##### fast check tests  #####
-
-- label: Documentation Build # 2min
-  working_dir: "/vllm-workspace/test_docs/docs"
-  fast_check: true
-  no_gpu: True
-  commands:
-  - pip install -r requirements-docs.txt
-  - SPHINXOPTS=\"-W\" make html
-  # Check API reference (if it fails, you may have missing mock imports)
-  - grep \"sig sig-object py\" build/html/dev/sampling_params.html
-
-- label: Async Engine, Inputs, Utils, Worker Test # 15min
-  fast_check: true
-  source_file_dependencies:
-  - vllm/
-  - tests/mq_llm_engine
-  - tests/async_engine
-  - tests/test_inputs
-  - tests/multimodal
-  - tests/test_utils
-  - tests/worker
-  commands:
-  - pytest -v -s mq_llm_engine # MQLLMEngine
-  - pytest -v -s async_engine # AsyncLLMEngine
-  - NUM_SCHEDULER_STEPS=4 pytest -v -s async_engine/test_async_llm_engine.py
-  - pytest -v -s test_inputs.py
-  - pytest -v -s multimodal
-  - pytest -v -s test_utils.py # Utils
-  - pytest -v -s worker # Worker
-
-- label: Basic Correctness Test # 30min
-  #mirror_hardwares: [amd]
-  fast_check: true
-  source_file_dependencies:
-  - vllm/
-  - tests/basic_correctness
-  commands:
-  - pytest -v -s basic_correctness/test_basic_correctness.py
-  - pytest -v -s basic_correctness/test_cpu_offload.py
-  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_chunked_prefill.py
-  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_chunked_prefill.py
-  - VLLM_TEST_ENABLE_ARTIFICIAL_PREEMPT=1 pytest -v -s basic_correctness/test_preemption.py
-
-- label: Core Test # 10min
-  mirror_hardwares: [amd]
-  fast_check: true
-  source_file_dependencies:
-  - vllm/core
-  - vllm/distributed
-  - tests/core
-  commands:
-  - pytest -v -s core
-
-- label: Entrypoints Test # 20min
-  working_dir: "/vllm-workspace/tests"
-  fast_check: true
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  commands:
-  - pip install -e ./plugins/vllm_add_dummy_model
-  - pip install git+https://github.com/EleutherAI/lm-evaluation-harness.git@a4987bba6e9e9b3f22bd3a6c1ecf0abd04fd5622#egg=lm_eval[api]
-  - pytest -v -s entrypoints/llm --ignore=entrypoints/llm/test_lazy_outlines.py --ignore=entrypoints/llm/test_generate.py --ignore=entrypoints/llm/test_generate_multiple_loras.py --ignore=entrypoints/llm/test_guided_generate.py
-  - pytest -v -s entrypoints/llm/test_lazy_outlines.py # it needs a clean process
-  - pytest -v -s entrypoints/llm/test_generate.py # it needs a clean process
-  - pytest -v -s entrypoints/llm/test_generate_multiple_loras.py # it needs a clean process
-  - pytest -v -s entrypoints/llm/test_guided_generate.py # it needs a clean process
-  - pytest -v -s entrypoints/openai
-  - pytest -v -s entrypoints/test_chat_utils.py
-  - pytest -v -s entrypoints/offline_mode # Needs to avoid interference with other tests
-
-- label: Distributed Tests (4 GPUs) # 10min
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 4
-  fast_check: true
-  source_file_dependencies:
-  - vllm/distributed/
-  - vllm/core/
-  - tests/distributed
-  - tests/spec_decode/e2e/test_integration_dist_tp4
-  commands:
-  - pytest -v -s distributed/test_pynccl.py
-  - pytest -v -s spec_decode/e2e/test_integration_dist_tp4.py
-
-- label: Metrics, Tracing Test # 10min
-  num_gpus: 2 
-  fast_check: true
-  source_file_dependencies:
-  - vllm/
-  - tests/metrics
-  - tests/tracing
-  commands:
-  - pytest -v -s metrics 
-  - "pip install \
-      'opentelemetry-sdk>=1.26.0,<1.27.0' \
-      'opentelemetry-api>=1.26.0,<1.27.0' \
-      'opentelemetry-exporter-otlp>=1.26.0,<1.27.0' \
-      'opentelemetry-semantic-conventions-ai>=0.4.1,<0.5.0'"
-  - pytest -v -s tracing
-
-##### fast check tests  #####
-#####  1 GPU test  #####
-
-- label: Regression Test # 5min
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  - tests/test_regression
+- label: Regression Test
   command: pytest -v -s test_regression.py
   working_dir: "/vllm-workspace/tests" # optional
 
-- label: Engine Test # 10min
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  - tests/engine
-  - tests/tokenization
-  commands:
-  - pytest -v -s engine test_sequence.py test_config.py test_logger.py
-  # OOM in the CI unless we run this separately
-  - pytest -v -s tokenization
+- label: AsyncEngine Test
+  command: pytest -v -s async_engine
 
-- label: Examples Test # 12min
-  working_dir: "/vllm-workspace/examples"
-  #mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/entrypoints
-  - examples/
+- label: Basic Correctness Test
   commands:
-    - pip install awscli tensorizer # for llava example and tensorizer test
+  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_basic_correctness.py
+  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_basic_correctness.py
+  - VLLM_ATTENTION_BACKEND=ROCM_FLASH pytest -v -s basic_correctness/test_basic_correctness.py
+  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_chunked_prefill.py
+  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_chunked_prefill.py
+  - VLLM_ATTENTION_BACKEND=ROCM_FLASH pytest -v -s basic_correctness/test_chunked_prefill.py
+
+- label: Core Test
+  command: pytest -v -s core
+
+- label: Distributed Comm Ops Test
+  command: pytest -v -s test_comm_ops.py
+  working_dir: "/vllm-workspace/tests/distributed"
+  num_gpus: 2 # only support 1 or 2 for now.
+
+- label: Distributed Tests
+  working_dir: "/vllm-workspace/tests/distributed"
+  num_gpus: 2 # only support 1 or 2 for now.
+  commands:
+  - pytest -v -s test_pynccl.py
+  - TEST_DIST_MODEL=facebook/opt-125m pytest -v -s test_basic_distributed_correctness.py
+  - TEST_DIST_MODEL=meta-llama/Llama-2-7b-hf pytest -v -s test_basic_distributed_correctness.py
+  - TEST_DIST_MODEL=facebook/opt-125m pytest -v -s test_chunked_prefill_distributed.py
+  - TEST_DIST_MODEL=meta-llama/Llama-2-7b-hf pytest -v -s test_chunked_prefill_distributed.py
+
+- label: Engine Test
+  command: pytest -v -s engine tokenization test_sequence.py test_config.py
+
+- label: Entrypoints Test
+  commands:
+  # these tests have to be separated, because each one will allocate all posible GPU memory
+  - pytest -v -s entrypoints --ignore=entrypoints/test_server_oot_registration.py
+  - pytest -v -s entrypoints/test_server_oot_registration.py
+
+- label: Examples Test
+  working_dir: "/vllm-workspace/examples"
+  commands:
+    # install aws cli for llava_example.py
+    - pip install awscli
     - python3 offline_inference.py
-    - python3 cpu_offload.py
-    - python3 offline_inference_chat.py
     - python3 offline_inference_with_prefix.py
     - python3 llm_engine_example.py
-    - python3 offline_inference_vision_language.py
-    - python3 offline_inference_vision_language_multi_image.py
-    - python3 tensorize_vllm_model.py --model facebook/opt-125m serialize --serialized-directory /tmp/ --suffix v1 && python3 tensorize_vllm_model.py --model facebook/opt-125m deserialize --path-to-tensors /tmp/vllm/facebook/opt-125m/v1/model.tensors
-    - python3 offline_inference_encoder_decoder.py
+    - python3 llava_example.py
 
-- label: Prefix Caching Test # 7min
-  #mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  - tests/prefix_caching
+- label: Kernels Test %N
+  command: pytest -v -s kernels --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
+  parallelism: 4
+
+- label: Models Test
+  commands:
+    - bash ../.buildkite/download-images.sh
+    - pytest -v -s models --ignore=models/test_llava.py --ignore=models/test_mistral.py
+
+- label: Llava Test
+  commands:
+    - bash ../.buildkite/download-images.sh
+    - pytest -v -s models/test_llava.py
+
+- label: Prefix Caching Test
   commands:
     - pytest -v -s prefix_caching
 
-- label: Samplers Test # 18min
-  source_file_dependencies:
-  - vllm/model_executor/layers
-  - vllm/sampling_metadata.py
-  - tests/samplers
-  commands:
-    - pytest -v -s samplers
-    - VLLM_USE_FLASHINFER_SAMPLER=1 pytest -v -s samplers
+- label: Samplers Test
+  command: pytest -v -s samplers
 
-- label: LogitsProcessor Test # 5min
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/model_executor/layers
-  - tests/test_logits_processor
+- label: LogitsProcessor Test
   command: pytest -v -s test_logits_processor.py
 
-- label: Speculative decoding tests # 22min
-  source_file_dependencies:
-  - vllm/spec_decode
-  - tests/spec_decode
-  commands:
-    # See https://github.com/vllm-project/vllm/issues/5152
-    - export VLLM_ATTENTION_BACKEND=XFORMERS
-    - pytest -v -s spec_decode/e2e/test_multistep_correctness.py
-    - pytest -v -s spec_decode --ignore=spec_decode/e2e/test_multistep_correctness.py
+- label: Worker Test
+  command: pytest -v -s worker
 
-- label: LoRA Test %N # 30min each
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/lora
-  - tests/lora
-  command: pytest -v -s lora --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --ignore=lora/test_long_context.py
+- label: Speculative decoding tests
+  command: pytest -v -s spec_decode
+
+- label: LoRA Test %N
+  command: pytest -v -s lora --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
   parallelism: 4
 
-- label: "PyTorch Fullgraph Smoke Test"
-  fast_check: true
-  source_file_dependencies:
-  - vllm/
-  - tests/compile
-  commands:
-  - pytest -v -s compile/test_full_graph_smoke.py
+- label: Tensorizer Test
+  command: apt-get install curl libsodium23 && pytest -v -s tensorizer
 
-- label: "PyTorch Fullgraph Test"
-  source_file_dependencies:
-  - vllm/
-  - tests/compile
-  commands:
-  - pytest -v -s compile/test_full_graph.py
+- label: Metrics Test
+  command: pytest -v -s metrics
 
-- label: Kernels Test %N # 30min each
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - csrc/
-  - vllm/attention
-  - tests/kernels
-  commands:
-    - pytest -v -s kernels --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
-  parallelism: 4
-
-- label: Tensorizer Test # 11min
-  mirror_hardwares: [amd]
-  soft_fail: true
-  source_file_dependencies:
-  - vllm/model_executor/model_loader
-  - tests/tensorizer_loader
-  commands:
-    - apt-get update && apt-get install -y curl libsodium23
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -v -s tensorizer_loader
-
-- label: Benchmarks # 9min
+- label: Benchmarks
   working_dir: "/vllm-workspace/.buildkite"
-  mirror_hardwares: [amd]
-  source_file_dependencies:
-  - benchmarks/
   commands:
   - pip install aiohttp
   - bash run-benchmarks.sh
 
-- label: Quantization Test # 15min
-  source_file_dependencies:
-  - csrc/
-  - vllm/model_executor/layers/quantization
-  - tests/quantization
-  command: pytest -v -s quantization
-
-- label: LM Eval Small Models # 53min
-  working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
-  source_file_dependencies:
-  - csrc/
-  - vllm/model_executor/layers/quantization
+- label: Documentation Build
+  working_dir: "/vllm-workspace/test_docs/docs"
+  no_gpu: True
   commands:
-  - pip install lm-eval
-  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-  - bash ./run-tests.sh -c configs/models-small.txt -t 1
-
-- label: Encoder Decoder tests # 5min
-  source_file_dependencies:
-  - vllm/
-  - tests/encoder_decoder
-  commands:
-    - pytest -v -s encoder_decoder
-
-- label: OpenAI-Compatible Tool Use # 20 min
-  fast_check: false
-  mirror_hardwares: [ amd ]
-  source_file_dependencies:
-    - vllm/
-    - tests/tool_use
-  commands:
-    - pytest -v -s tool_use
-
-#####  models test  #####
-
-- label: Basic Models Test # 3min
-  source_file_dependencies:
-  - vllm/
-  - tests/models
-  commands:
-    - pip install -e ./plugins/vllm_add_dummy_model
-    - pytest -v -s models/test_oot_registration.py # it needs a clean process
-    - pytest -v -s models/*.py --ignore=models/test_oot_registration.py
-
-- label: Decoder-only Language Models Test # 1h3min
-  #mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  - tests/models/decoder_only/language
-  commands:
-    - pytest -v -s models/decoder_only/language
-
-- label: Decoder-only Multi-Modal Models Test # 56min
-  #mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  - tests/models/decoder_only/audio_language
-  - tests/models/decoder_only/vision_language
-  commands:
-    - pytest -v -s models/decoder_only/audio_language
-    - pytest -v -s models/decoder_only/vision_language
-
-- label: Other Models Test # 5min
-  #mirror_hardwares: [amd]
-  source_file_dependencies:
-  - vllm/
-  - tests/models/embedding/language
-  - tests/models/encoder_decoder/language
-  commands:
-    - pytest -v -s models/embedding/language
-    - pytest -v -s models/encoder_decoder/language
-
-#####  1 GPU test  #####
-#####  multi gpus test  #####
-
-- label: Distributed Comm Ops Test # 7min
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 2
-  source_file_dependencies:
-  - vllm/distributed
-  - tests/distributed
-  commands:
-  - pytest -v -s distributed/test_comm_ops.py
-  - pytest -v -s distributed/test_shm_broadcast.py
-
-- label: 2 Node Tests (4 GPUs in total) # 16min
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 2
-  num_nodes: 2
-  source_file_dependencies:
-  - vllm/distributed/
-  - vllm/engine/
-  - vllm/executor/
-  - vllm/model_executor/models/
-  - tests/distributed/
-  commands:
-  - # the following commands are for the first node, with ip 192.168.10.10 (ray environment already set up)
-    - VLLM_TEST_SAME_HOST=0 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_same_node.py | grep -q 'Same node test passed'
-    - VLLM_MULTI_NODE=1 pytest -v -s distributed/test_multi_node_assignment.py
-    - VLLM_MULTI_NODE=1 pytest -v -s distributed/test_pipeline_parallel.py
-  - # the following commands are for the second node, with ip 192.168.10.11 (ray environment already set up)
-    - VLLM_TEST_SAME_HOST=0 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_same_node.py | grep -q 'Same node test passed'
-
-- label: Distributed Tests (2 GPUs) # 28min
-  #mirror_hardwares: [amd]
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 2
-  source_file_dependencies:
-  - vllm/distributed/
-  - vllm/engine/
-  - vllm/executor/
-  - vllm/model_executor/models/
-  - tests/distributed/
-  - vllm/compilation
-  commands:
-  - pytest -v -s ./compile/test_full_graph_multi_gpu.py
-  - pytest -v -s ./compile/test_wrapper.py
-  - VLLM_TEST_SAME_HOST=1 torchrun --nproc-per-node=4 distributed/test_same_node.py | grep -q 'Same node test passed'
-  - TARGET_TEST_SUITE=L4 pytest basic_correctness/ -v -s -m distributed_2_gpus
-  # Avoid importing model tests that cause CUDA reinitialization error
-  - pytest models/encoder_decoder/language/test_bart.py models/decoder_only/vision_language/test_broadcast.py -v -s -m distributed_2_gpus
-  - pytest -v -s spec_decode/e2e/test_integration_dist_tp2.py
-  - pip install -e ./plugins/vllm_add_dummy_model
-  - pytest -v -s distributed/test_distributed_oot.py
-  - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s test_sharded_state_loader.py
-  - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s distributed/test_utils.py
-
-- label: Multi-step Tests (4 GPUs) # 21min
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 4
-  source_file_dependencies:
-  - vllm/model_executor/layers/sampler.py
-  - vllm/sequence.py
-  - vllm/worker/worker_base.py
-  - vllm/worker/worker.py
-  - vllm/worker/multi_step_worker.py
-  - vllm/worker/model_runner_base.py
-  - vllm/worker/model_runner.py
-  - vllm/worker/multi_step_model_runner.py
-  - vllm/engine
-  - tests/multi_step
-  commands:
-  - pytest -v -s multi_step/test_correctness_async_llm.py
-  - pytest -v -s multi_step/test_correctness_llm.py
-
-- label: Pipeline Parallelism Test # 23min
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 4
-  source_file_dependencies:
-  - vllm/distributed/
-  - vllm/engine/
-  - vllm/executor/
-  - vllm/model_executor/models/
-  - tests/distributed/
-  commands:
-  - pytest -v -s distributed/test_pp_cudagraph.py
-  - pytest -v -s distributed/test_pipeline_parallel.py
-
-- label: LoRA Long Context (Distributed) # 11min
-  # This test runs llama 13B, so it is required to run on 4 GPUs.
-  num_gpus: 4
-  soft_fail: true
-  source_file_dependencies:
-  - vllm/lora
-  - tests/lora/test_long_context
-  commands:
-    # FIXIT: find out which code initialize cuda before running the test
-    # before the fix, we need to use spawn to test it
-    - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-    - pytest -v -s -x lora/test_long_context.py
-
-- label: Weight Loading Multiple GPU Test
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 2
-  source_file_dependencies:
-  - vllm/
-  - tests/weight_loading
-  commands:
-    - bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models.txt
-
-- label: Weight Loading Multiple GPU Test - Large Models # optional
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 2
-  gpu: a100
-  optional: true
-  source_file_dependencies:
-  - vllm/
-  - tests/weight_loading
-  commands:
-    - bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models-large.txt 
-
-
-##### multi gpus test #####
-##### A100 test #####
-
-- label: Distributed Tests (A100) # optional
-  gpu: a100
-  num_gpus: 4
-  source_file_dependencies:
-  - vllm/
-  commands: 
-  # NOTE: don't test llama model here, it seems hf implementation is buggy
-  # see https://github.com/vllm-project/vllm/pull/5689 for details
-  - pytest -v -s distributed/test_custom_all_reduce.py
-  - TARGET_TEST_SUITE=A100 pytest -v -s distributed/test_basic_distributed_correctness.py
-  - pytest -v -s -x lora/test_mixtral.py
-
-- label: LM Eval Large Models # optional
-  gpu: a100
-  num_gpus: 4
-  working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
-  source_file_dependencies:
-  - csrc/
-  - vllm/model_executor/layers/quantization
-  commands:
-  - pip install lm-eval
-  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-  - bash ./run-tests.sh -c configs/models-large.txt -t 4
+  - pip install -r requirements-docs.txt
+  - SPHINXOPTS=\"-W\" make html

.buildkite/test-template.j2

@@ -0,0 +1,69 @@
+{% set docker_image = "us-central1-docker.pkg.dev/vllm-405802/vllm-ci-test-repo/vllm-test:$BUILDKITE_COMMIT" %}
+{% set default_num_gpu = 1 %}
+{% set default_working_dir = "/vllm-workspace/tests" %}
+
+steps:
+  - label: "AMD Test"
+    agents:
+      queue: amd
+    command: bash .buildkite/run-amd-test.sh
+
+  - label: "CPU Test"
+    command: bash .buildkite/run-cpu-test.sh
+
+  - label: ":docker: build image"
+    commands:
+      - "docker build --build-arg max_jobs=16 --tag {{ docker_image }} --target test --progress plain ."
+      - "docker push {{ docker_image }}"
+    env:
+      DOCKER_BUILDKIT: "1"
+    retry:
+      automatic:
+        - exit_status: -1  # Agent was lost
+          limit: 5
+  - wait
+
+  {% for step in steps %}
+  - label: "{{ step.label }}"
+    agents:
+      queue: kubernetes
+    soft_fail: {{ step.soft_fail or false }}
+    {% if step.parallelism %}
+    parallelism: {{ step.parallelism }}
+    {% endif %}
+    retry:
+      automatic:
+        - exit_status: -1  # Agent was lost
+          limit: 5
+    plugins:
+      - kubernetes:
+          podSpec:
+            volumes:
+              - name: dshm
+                emptyDir:
+                  medium: Memory
+            containers:
+              - image: "{{ docker_image }}"
+                command: ["bash"]
+                args:
+                - '-c'
+                - "'cd {{ (step.working_dir or default_working_dir) | safe  }} && {{ step.command  or (step.commands | join(' && ')) | safe }}'"
+                {% if not step.no_gpu %}
+                resources:
+                  requests:
+                    nvidia.com/gpu: "{{ step.num_gpus or default_num_gpu }}"
+                  limits:
+                    nvidia.com/gpu: "{{ step.num_gpus or default_num_gpu }}"
+                {% endif %}
+                env:
+                  - name: VLLM_USAGE_SOURCE
+                    value: ci-test
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
+                volumeMounts:
+                  - mountPath: /dev/shm
+                    name: dshm
+  {% endfor %}

.clang-format

@@ -1,26 +0,0 @@
-BasedOnStyle: Google
-UseTab: Never
-IndentWidth: 2
-ColumnLimit: 80
-
-# Force pointers to the type for C++.
-DerivePointerAlignment: false
-PointerAlignment: Left
-
-# Reordering #include statements can (and currently will) introduce errors
-SortIncludes: false
-
-# Style choices
-AlignConsecutiveAssignments: false
-AlignConsecutiveDeclarations: false
-IndentPPDirectives: BeforeHash
-
-IncludeCategories:
-  - Regex:           '^<'
-    Priority:        4
-  - Regex:           '^"(llvm|llvm-c|clang|clang-c|mlir|mlir-c)/'
-    Priority:        3
-  - Regex:           '^"(qoda|\.\.)/'
-    Priority:        2
-  - Regex:           '.*'
-    Priority:        1

.dockerignore

@@ -1,4 +1 @@
 vllm/*.so
-/.venv
-/build
-dist

.github/FUNDING.yml

@@ -1,2 +0,0 @@
-github: [vllm-project]
-open_collective: [vllm]

.github/ISSUE_TEMPLATE/100-documentation.yml

@@ -20,10 +20,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/200-installation.yml

@@ -18,7 +18,6 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
-      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`
@@ -38,10 +37,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/300-usage.yml

@@ -18,7 +18,6 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
-      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`
@@ -36,10 +35,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/400-bug report.yml

@@ -18,27 +18,12 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
-      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
-      <details>
-      <summary>The output of `python collect_env.py`</summary>
-
       ```text
-      Your output of `python collect_env.py` here
+      The output of `python collect_env.py`
       ```
-      
-      </details>
   validations:
     required: true
-- type: textarea
-  attributes:
-    label: Model Input Dumps
-    description: |
-      If you are facing crashing due to illegal memory access or other issues with model execution, vLLM may dump the problematic input of the model. In this case, you will see the message `Error in model execution (input dumped to /tmp/err_xxx.pkl)`. If you see this message, please zip the file (because GitHub doesn't support .pkl file format) and upload it here. This will help us to reproduce the issue and facilitate the debugging process.
-    placeholder: |
-      Upload the dumped input file.
-  validations:
-    required: false
 - type: textarea
   attributes:
     label: 🐛 Describe the bug
@@ -72,10 +57,6 @@ body:
       If the code is too long (hopefully, it isn't), feel free to put it in a public gist and link it in the issue: https://gist.github.com.
 
       Please also paste or describe the results you observe instead of the expected results. If you observe an error, please paste the error message including the **full** traceback of the exception. It may be relevant to wrap error messages in ```` ```triple quotes blocks``` ````.
-
-      Please set the environment variable `export VLLM_LOGGING_LEVEL=DEBUG` to turn on more logging to help debugging potential issues.
-
-      If you experienced crashes or hangs, it would be helpful to run vllm with `export VLLM_TRACE_FUNCTION=1` . All the function calls in vllm will be recorded. Inspect these log files, and tell which function crashes or hangs.
     placeholder: |
       A clear and concise description of what the bug is.
 
@@ -98,10 +79,3 @@ body:
       - If the error only appears in vllm, please provide the detailed script of how you run `transformers` and `vllm`, also highlight the difference and what you expect.
 
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/500-feature request.yml

@@ -29,10 +29,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/600-new model.yml

@@ -31,10 +31,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/700-performance discussion.yml

@@ -39,7 +39,6 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
-      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`
@@ -50,10 +49,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/750-RFC.yml

@@ -1,56 +0,0 @@
-name: 💬 Request for comments (RFC).
-description: Ask for feedback on major architectural changes or design choices.
-title: "[RFC]: "
-labels: ["RFC"]
-
-body:
-- type: markdown
-  attributes:
-    value: >
-      #### Please take a look at previous [RFCs](https://github.com/vllm-project/vllm/issues?q=label%3ARFC+sort%3Aupdated-desc) for reference.
-- type: textarea
-  attributes:
-    label: Motivation.
-    description: >
-      The motivation of the RFC.
-  validations:
-    required: true
-- type: textarea
-  attributes:
-    label: Proposed Change.
-    description: >
-      The proposed change of the RFC.
-  validations:
-    required: true
-- type: textarea
-  attributes:
-    label: Feedback Period.
-    description: >
-      The feedback period of the RFC. Usually at least one week.
-  validations:
-    required: false
-- type: textarea
-  attributes:
-    label: CC List.
-    description: >
-      The list of people you want to CC.
-  validations:
-    required: false
-- type: textarea
-  attributes:
-    label: Any Other Things.
-    description: >
-      Any other things you would like to mention.
-  validations:
-    required: false
-- type: markdown
-  attributes:
-    value: >
-      Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/ISSUE_TEMPLATE/800-misc discussion.yml

@@ -19,10 +19,3 @@ body:
   attributes:
     value: >
       Thanks for contributing 🎉!
-- type: checkboxes
-  id: askllm
-  attributes:
-    label: Before submitting a new issue...
-    options:
-      - label: Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the [documentation page](https://docs.vllm.ai/en/latest/), which can answer lots of frequently asked questions.
-        required: true

.github/PULL_REQUEST_TEMPLATE.md

@@ -39,16 +39,6 @@ FIX #xxxx (*link existing issues this PR will resolve*)
     <li>Please add documentation to <code>docs/source/</code> if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.</li>
 </ul>
 
-<h3>Adding or changing kernels</h3>
-<p>Each custom kernel needs a schema and one or more implementations to be registered with PyTorch.</p>
-<ul>
-    <li>Make sure custom ops are registered following PyTorch guidelines: <a href="https://pytorch.org/tutorials/advanced/cpp_custom_ops.html#cpp-custom-ops-tutorial">Custom C++ and CUDA Operators</a> and <a href="https://docs.google.com/document/d/1_W62p8WJOQQUzPsJYa7s701JXt0qf2OfLub2sbkHOaU">The Custom Operators Manual</a></li>
-    <li>Custom operations that return <code>Tensors</code> require meta-functions. Meta-functions should be implemented and registered in python so that dynamic dims can be handled automatically. See above documents for a description of meta-functions.</li>
-    <li>Use <a href="https://pytorch.org/docs/stable/library.html#torch.library.opcheck"><code>torch.libary.opcheck()</code></a> to test the function registration and meta-function for any registered ops.  See <code>tests/kernels</code> for examples.</li>
-    <li>When changing the C++ signature of an existing op, the schema must be updated to reflect the changes.</li>
-    <li>If a new custom type is needed, see the following document: <a href="https://docs.google.com/document/d/18fBMPuOJ0fY5ZQ6YyrHUppw9FA332CpNtgB6SOIgyuA">Custom Class Support in PT2</a>.
-</ul>
-
 <h3>Notes for Large Changes</h3>
 <p>Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with <code>rfc-required</code> and might not go through the PR.</p>
 

.github/workflows/add_label_automerge.yml

@@ -1,21 +0,0 @@
-name: Add label on auto-merge enabled
-on:
-    pull_request_target:
-        types:
-            - auto_merge_enabled
-jobs:
-    add-label-on-auto-merge:
-        runs-on: ubuntu-latest
-        steps:
-            -   name: Add label
-                uses: actions/github-script@v5
-                with:
-                    script: |
-                        github.rest.issues.addLabels({
-                            owner: context.repo.owner,
-                            repo: context.repo.repo,
-                            issue_number: context.issue.number,
-                            labels: ['ready']
-                        })
-                env:
-                    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/clang-format.yml

@@ -1,41 +0,0 @@
-name: clang-format
-
-on:
-  # Trigger the workflow on push or pull request,
-  # but only for the main branch
-  push:
-    branches:
-      - main
-  pull_request:
-    branches:
-      - main
-
-jobs:
-  clang-format:
-    runs-on: ubuntu-latest
-    strategy:
-      matrix:
-        python-version: ["3.11"]
-    steps:
-    - uses: actions/checkout@v2
-    - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v2
-      with:
-        python-version: ${{ matrix.python-version }}
-    - name: Install dependencies
-      run: |
-        python -m pip install --upgrade pip
-        pip install clang-format==18.1.5
-    - name: Running clang-format
-      run: |
-        EXCLUDES=(
-            'csrc/moe/topk_softmax_kernels.cu'
-            'csrc/quantization/gguf/ggml-common.h'
-            'csrc/quantization/gguf/dequantize.cuh'
-            'csrc/quantization/gguf/vecdotq.cuh'
-            'csrc/quantization/gguf/mmq.cuh'
-            'csrc/quantization/gguf/mmvq.cuh'
-        )
-        find csrc/ \( -name '*.h' -o -name '*.cpp' -o -name '*.cu' -o -name '*.cuh' \) -print \
-            | grep -vFf <(printf "%s\n" "${EXCLUDES[@]}") \
-            | xargs clang-format --dry-run --Werror

.github/workflows/mypy.yaml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}
@@ -25,22 +25,26 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install mypy==1.11.1
+        pip install mypy==1.9.0
         pip install types-setuptools
         pip install types-PyYAML
         pip install types-requests
         pip install types-setuptools
     - name: Mypy
       run: |
-        mypy
-        mypy tests --follow-imports skip
-        mypy vllm/attention --follow-imports skip
-        mypy vllm/distributed --follow-imports skip
-        mypy vllm/engine  --follow-imports skip
-        mypy vllm/executor --follow-imports skip
-        mypy vllm/lora --follow-imports skip
-        mypy vllm/model_executor  --follow-imports skip
-        mypy vllm/prompt_adapter --follow-imports skip
-        mypy vllm/spec_decode --follow-imports skip
-        mypy vllm/worker --follow-imports skip
+        mypy vllm/attention/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/core/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/distributed/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/entrypoints/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/executor/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/usage/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/transformers_utils/*.py --follow-imports=skip --config-file pyproject.toml
+
+        # TODO(sang): Follow up
+        # mypy vllm/engine/*.py --follow-imports=skip --config-file pyproject.toml
+        # mypy vllm/worker/*.py --follow-imports=skip --config-file pyproject.toml
+        # mypy vllm/spec_decoding/*.py --follow-imports=skip --config-file pyproject.toml
+        # mypy vllm/model_executor/*.py --follow-imports=skip --config-file pyproject.toml
+        # mypy vllm/lora/*.py --follow-imports=skip --config-file pyproject.toml
 

.github/workflows/publish.yml

@@ -48,20 +48,14 @@ jobs:
       fail-fast: false
       matrix:
           os: ['ubuntu-20.04']
-          python-version: ['3.8', '3.9', '3.10', '3.11', '3.12']
-          pytorch-version: ['2.4.0']  # Must be the most recent version that meets requirements-cuda.txt.
+          python-version: ['3.8', '3.9', '3.10', '3.11']
+          pytorch-version: ['2.2.1']  # Must be the most recent version that meets requirements-cuda.txt.
           cuda-version: ['11.8', '12.1']
 
     steps:
       - name: Checkout
         uses: actions/checkout@v3
 
-      - name: Setup ccache
-        uses: hendrikmuhs/ccache-action@v1.2
-        with:
-          create-symlink: true
-          key: ${{ github.job }}-${{ matrix.python-version }}-${{ matrix.cuda-version }}
-
       - name: Set up Linux Env
         if: ${{ runner.os == 'Linux' }}
         run: |
@@ -82,8 +76,6 @@ jobs:
 
       - name: Build wheel
         shell: bash
-        env:
-          CMAKE_BUILD_TYPE: Release # do not compile with debug symbol to reduce wheel size
         run: |
           bash -x .github/workflows/scripts/build.sh ${{ matrix.python-version }} ${{ matrix.cuda-version }}
           wheel_name=$(ls dist/*whl | xargs -n 1 basename)

.github/workflows/reminder_comment.yml

@@ -1,21 +0,0 @@
-name: PR Reminder Comment Bot
-on:
-  pull_request_target:
-    types: [opened]
-
-jobs:
-  pr_reminder:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Remind to run full CI on PR
-        uses: actions/github-script@v6
-        with:
-          script: |
-            github.rest.issues.createComment({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              issue_number: context.issue.number,
-              body: '👋 Hi! Thank you for contributing to the vLLM project.\n Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run `fastcheck` CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your `fastcheck` build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping `simon-mo` or `khluu` to add you in our Buildkite org. \n\nOnce the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.\n\n To run CI, PR reviewers can do one of these:\n- Add `ready` label to the PR\n- Enable auto-merge.\n\n🚀'
-            })
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/ruff.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}
@@ -25,10 +25,10 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install -r requirements-lint.txt
+        pip install ruff==0.1.5 codespell==2.2.6 tomli==2.0.1 isort==5.13.2
     - name: Analysing the code with ruff
       run: |
-        ruff check .
+        ruff .
     - name: Spelling check with codespell
       run: |
         codespell --toml pyproject.toml

.github/workflows/scripts/build.sh

@@ -13,8 +13,9 @@ $python_executable -m pip install -r requirements-cuda.txt
 
 # Limit the number of parallel jobs to avoid OOM
 export MAX_JOBS=1
+# Make sure punica is built for the release (for LoRA)
+export VLLM_INSTALL_PUNICA_KERNELS=1
 # Make sure release wheels are built for the following architectures
 export TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 8.9 9.0+PTX"
-export VLLM_FA_CMAKE_GPU_ARCHES="80-real;90-real"
 # Build
 $python_executable setup.py bdist_wheel --dist-dir=dist

.github/workflows/scripts/create_release.js

@@ -8,7 +8,7 @@ module.exports = async (github, context, core) => {
 			generate_release_notes: true,
 			name: process.env.RELEASE_TAG,
 			owner: context.repo.owner,
-			prerelease: true,
+			prerelease: false,
 			repo: context.repo.repo,
 			tag_name: process.env.RELEASE_TAG,
 		});

.github/workflows/yapf.yml

@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}

.gitignore

@@ -1,9 +1,3 @@
-# version file generated by setuptools-scm
-/vllm/_version.py
-
-# vllm-flash-attn built from source
-vllm/vllm_flash_attn/
-
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]
@@ -15,8 +9,6 @@ __pycache__/
 # Distribution / packaging
 .Python
 build/
-cmake-build-*/
-CMakeUserPresets.json
 develop-eggs/
 dist/
 downloads/
@@ -78,8 +70,6 @@ instance/
 
 # Sphinx documentation
 docs/_build/
-docs/source/getting_started/examples/*.rst
-!**/*.template.rst
 
 # PyBuilder
 .pybuilder/
@@ -92,9 +82,6 @@ target/
 profile_default/
 ipython_config.py
 
-# generated files
-**/generated/**
-
 # pyenv
 #   For a library or package, you might want to ignore these files since the code is
 #   intended to run in multiple environments; otherwise, check them in:
@@ -197,4 +184,4 @@ _build/
 hip_compat.h
 
 # Benchmark dataset
-benchmarks/*.json
+*.json

.readthedocs.yaml

@@ -10,7 +10,6 @@ build:
 
 sphinx:
    configuration: docs/source/conf.py
-   fail_on_warning: true
 
 # If using Sphinx, optionally build your docs in additional formats such as PDF
 formats:

CMakeLists.txt

@@ -1,37 +1,19 @@
-cmake_minimum_required(VERSION 3.26)
+cmake_minimum_required(VERSION 3.21)
 
-# When building directly using CMake, make sure you run the install step
-# (it places the .so files in the correct location).
-#
-# Example:
-# mkdir build && cd build
-# cmake -G Ninja -DVLLM_PYTHON_EXECUTABLE=`which python3` -DCMAKE_INSTALL_PREFIX=.. ..
-# cmake --build . --target install
-#
-# If you want to only build one target, make sure to install it manually:
-# cmake --build . --target _C
-# cmake --install . --component _C
 project(vllm_extensions LANGUAGES CXX)
 
-# CUDA by default, can be overridden by using -DVLLM_TARGET_DEVICE=... (used by setup.py)
-set(VLLM_TARGET_DEVICE "cuda" CACHE STRING "Target device backend for vLLM")
+option(VLLM_TARGET_DEVICE "Target device backend for vLLM" "cuda")
 
 message(STATUS "Build type: ${CMAKE_BUILD_TYPE}")
 message(STATUS "Target device: ${VLLM_TARGET_DEVICE}")
 
 include(${CMAKE_CURRENT_LIST_DIR}/cmake/utils.cmake)
 
-# Suppress potential warnings about unused manually-specified variables
-set(ignoreMe "${VLLM_PYTHON_PATH}")
-
-# Prevent installation of dependencies (cutlass) by default.
-install(CODE "set(CMAKE_INSTALL_LOCAL_ONLY TRUE)" ALL_COMPONENTS)
-
 #
 # Supported python versions.  These versions will be searched in order, the
 # first match will be selected.  These should be kept in sync with setup.py.
 #
-set(PYTHON_SUPPORTED_VERSIONS "3.8" "3.9" "3.10" "3.11" "3.12")
+set(PYTHON_SUPPORTED_VERSIONS "3.8" "3.9" "3.10" "3.11")
 
 # Supported NVIDIA architectures.
 set(CUDA_SUPPORTED_ARCHS "7.0;7.5;8.0;8.6;8.9;9.0")
@@ -49,8 +31,9 @@ set(HIP_SUPPORTED_ARCHS "gfx906;gfx908;gfx90a;gfx940;gfx941;gfx942;gfx1030;gfx11
 # requirements.txt files and should be kept consistent.  The ROCm torch
 # versions are derived from Dockerfile.rocm
 #
-set(TORCH_SUPPORTED_VERSION_CUDA "2.4.0")
-set(TORCH_SUPPORTED_VERSION_ROCM "2.5.0")
+set(TORCH_SUPPORTED_VERSION_CUDA "2.2.1")
+set(TORCH_SUPPORTED_VERSION_ROCM_5X "2.0.1")
+set(TORCH_SUPPORTED_VERSION_ROCM_6X "2.1.1")
 
 #
 # Try to find python package with an executable that exactly matches
@@ -84,22 +67,17 @@ endif()
 find_package(Torch REQUIRED)
 
 #
-message(STATUS "Enabling core extension.")
-
-# Define _core_C extension
-#  built for (almost) every target platform, (excludes TPU and Neuron)
-
-set(VLLM_EXT_SRC
-  "csrc/core/torch_bindings.cpp")
-
-define_gpu_extension_target(
-  _core_C
-  DESTINATION vllm
-  LANGUAGE CXX
-  SOURCES ${VLLM_EXT_SRC}
-  COMPILE_FLAGS ${CXX_COMPILE_FLAGS}
-  USE_SABI 3
-  WITH_SOABI)
+# Normally `torch.utils.cpp_extension.CUDAExtension` would add
+# `libtorch_python.so` for linking against an extension. Torch's cmake
+# configuration does not include this library (presumably since the cmake
+# config is used for standalone C++ binaries that link against torch).
+# The `libtorch_python.so` library defines some of the glue code between
+# torch/python via pybind and is required by VLLM extensions for this
+# reason. So, add it by manually with `find_library` using torch's
+# installed library path.
+#
+find_library(torch_python_LIBRARY torch_python PATHS
+  "${TORCH_INSTALL_PREFIX}/lib")
 
 #
 # Forward the non-CUDA device extensions to external CMake scripts.
@@ -109,7 +87,7 @@ if (NOT VLLM_TARGET_DEVICE STREQUAL "cuda" AND
     if (VLLM_TARGET_DEVICE STREQUAL "cpu")
         include(${CMAKE_CURRENT_LIST_DIR}/cmake/cpu_extension.cmake)
     else()
-        return()
+        message(FATAL_ERROR "Unsupported vLLM target device: ${VLLM_TARGET_DEVICE}")
     endif()
     return()
 endif()
@@ -133,11 +111,18 @@ elseif(HIP_FOUND)
   # .hip extension automatically, HIP must be enabled explicitly.
   enable_language(HIP)
 
-  # ROCm 5.X and 6.X
-  if (ROCM_VERSION_DEV_MAJOR GREATER_EQUAL 5 AND
-      NOT Torch_VERSION VERSION_EQUAL ${TORCH_SUPPORTED_VERSION_ROCM})
-    message(WARNING "Pytorch version >= ${TORCH_SUPPORTED_VERSION_ROCM} "
-      "expected for ROCm build, saw ${Torch_VERSION} instead.")
+  # ROCm 5.x
+  if (ROCM_VERSION_DEV_MAJOR EQUAL 5 AND
+      NOT Torch_VERSION VERSION_EQUAL ${TORCH_SUPPORTED_VERSION_ROCM_5X})
+    message(WARNING "Pytorch version ${TORCH_SUPPORTED_VERSION_ROCM_5X} "
+      "expected for ROCMm 5.x build, saw ${Torch_VERSION} instead.")
+  endif()
+
+  # ROCm 6.x
+  if (ROCM_VERSION_DEV_MAJOR EQUAL 6 AND
+      NOT Torch_VERSION VERSION_EQUAL ${TORCH_SUPPORTED_VERSION_ROCM_6X})
+    message(WARNING "Pytorch version ${TORCH_SUPPORTED_VERSION_ROCM_6X} "
+      "expected for ROCMm 6.x build, saw ${Torch_VERSION} instead.")
   endif()
 else()
   message(FATAL_ERROR "Can't find CUDA or HIP installation.")
@@ -166,10 +151,8 @@ if(NVCC_THREADS AND VLLM_GPU_LANG STREQUAL "CUDA")
   list(APPEND VLLM_GPU_FLAGS "--threads=${NVCC_THREADS}")
 endif()
 
-include(FetchContent)
-
 #
-# Define other extension targets
+# Define extension targets
 #
 
 #
@@ -182,113 +165,19 @@ set(VLLM_EXT_SRC
   "csrc/pos_encoding_kernels.cu"
   "csrc/activation_kernels.cu"
   "csrc/layernorm_kernels.cu"
+  "csrc/quantization/squeezellm/quant_cuda_kernel.cu"
   "csrc/quantization/gptq/q_gemm.cu"
-  "csrc/quantization/compressed_tensors/int8_quant_kernels.cu"
-  "csrc/quantization/fp8/common.cu"
   "csrc/cuda_utils_kernels.cu"
   "csrc/moe_align_block_size_kernels.cu"
-  "csrc/prepare_inputs/advance_step.cu"
-  "csrc/torch_bindings.cpp")
+  "csrc/pybind.cpp")
 
 if(VLLM_GPU_LANG STREQUAL "CUDA")
-  SET(CUTLASS_ENABLE_HEADERS_ONLY ON CACHE BOOL "Enable only the header library")
-
-  # Set CUTLASS_REVISION manually -- its revision detection doesn't work in this case.
-  set(CUTLASS_REVISION "v3.5.1" CACHE STRING "CUTLASS revision to use")
-
-  FetchContent_Declare(
-        cutlass
-        GIT_REPOSITORY https://github.com/nvidia/cutlass.git
-        GIT_TAG v3.5.1
-        GIT_PROGRESS TRUE
-
-        # Speed up CUTLASS download by retrieving only the specified GIT_TAG instead of the history.
-        # Important: If GIT_SHALLOW is enabled then GIT_TAG works only with branch names and tags.
-        # So if the GIT_TAG above is updated to a commit hash, GIT_SHALLOW must be set to FALSE
-        GIT_SHALLOW TRUE
-  )
-  FetchContent_MakeAvailable(cutlass)
-
   list(APPEND VLLM_EXT_SRC
-    "csrc/mamba/mamba_ssm/selective_scan_fwd.cu"
-    "csrc/mamba/causal_conv1d/causal_conv1d.cu"
-    "csrc/quantization/aqlm/gemm_kernels.cu"
     "csrc/quantization/awq/gemm_kernels.cu"
-    "csrc/quantization/marlin/dense/marlin_cuda_kernel.cu"
-    "csrc/quantization/marlin/sparse/marlin_24_cuda_kernel.cu"
-    "csrc/quantization/marlin/qqq/marlin_qqq_gemm_kernel.cu"
-    "csrc/quantization/gptq_marlin/gptq_marlin.cu"
-    "csrc/quantization/gptq_marlin/gptq_marlin_repack.cu"
-    "csrc/quantization/gptq_marlin/awq_marlin_repack.cu"
-    "csrc/quantization/gguf/gguf_kernel.cu"
-    "csrc/quantization/fp8/fp8_marlin.cu"
-    "csrc/custom_all_reduce.cu"
-    "csrc/permute_cols.cu"
-    "csrc/quantization/cutlass_w8a8/scaled_mm_entry.cu"
-    "csrc/quantization/cutlass_w8a8/scaled_mm_c2x.cu"
-    "csrc/quantization/cutlass_w8a8/scaled_mm_c3x.cu")
-
-  #
-  # The CUTLASS kernels for Hopper require sm90a to be enabled.
-  # This is done via the below gencode option, BUT that creates kernels for both sm90 and sm90a.
-  # That adds an extra 17MB to compiled binary, so instead we selectively enable it.
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.0)
-    set_source_files_properties(
-          "csrc/quantization/cutlass_w8a8/scaled_mm_c3x.cu"
-          PROPERTIES
-          COMPILE_FLAGS
-          "-gencode arch=compute_90a,code=sm_90a")
-  endif()
-
-
-  #
-  # Machete kernels
-
-  # The machete kernels only work on hopper and require CUDA 12.0 or later.
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.0)
-    #
-    # For the Machete kernels we automatically generate sources for various 
-    # preselected input type pairs and schedules.
-    # Generate sources:
-    execute_process(
-      COMMAND ${CMAKE_COMMAND} -E env 
-      PYTHONPATH=${CMAKE_CURRENT_SOURCE_DIR}/csrc/cutlass_extensions/:${CUTLASS_DIR}/python/:${VLLM_PYTHON_PATH}:$PYTHONPATH 
-        ${Python_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/csrc/quantization/machete/generate.py
-      RESULT_VARIABLE machete_generation_result
-      OUTPUT_VARIABLE machete_generation_output
-      OUTPUT_FILE ${CMAKE_CURRENT_BINARY_DIR}/machete_generation.log
-      ERROR_FILE ${CMAKE_CURRENT_BINARY_DIR}/machete_generation.log
-    )
-
-    if (NOT machete_generation_result EQUAL 0)
-      message(FATAL_ERROR "Machete generation failed."
-                          " Result: \"${machete_generation_result}\"" 
-                          "\nCheck the log for details: "
-                          "${CMAKE_CURRENT_BINARY_DIR}/machete_generation.log")
-    else()
-      message(STATUS "Machete generation completed successfully.")
-    endif()
-
-    # Add machete generated sources
-    file(GLOB MACHETE_GEN_SOURCES "csrc/quantization/machete/generated/*.cu")
-    list(APPEND VLLM_EXT_SRC ${MACHETE_GEN_SOURCES})
-    message(STATUS "Machete generated sources: ${MACHETE_GEN_SOURCES}")
-
-    set_source_files_properties(
-          ${MACHETE_GEN_SOURCES}
-          PROPERTIES
-          COMPILE_FLAGS
-          "-gencode arch=compute_90a,code=sm_90a")
-  endif()
-
-  # Add pytorch binding for machete (add on even CUDA < 12.0 so that we can
-  #  raise an error if the user that this was built with an incompatible 
-  #  CUDA version)
-  list(APPEND VLLM_EXT_SRC
-    csrc/quantization/machete/machete_pytorch.cu)
+    "csrc/quantization/marlin/marlin_cuda_kernel.cu"
+    "csrc/custom_all_reduce.cu")
 endif()
 
-message(STATUS "Enabling C extension.")
 define_gpu_extension_target(
   _C
   DESTINATION vllm
@@ -296,35 +185,16 @@ define_gpu_extension_target(
   SOURCES ${VLLM_EXT_SRC}
   COMPILE_FLAGS ${VLLM_GPU_FLAGS}
   ARCHITECTURES ${VLLM_GPU_ARCHES}
-  INCLUDE_DIRECTORIES ${CUTLASS_INCLUDE_DIR}
-  USE_SABI 3
   WITH_SOABI)
 
-# If CUTLASS is compiled on NVCC >= 12.5, it by default uses 
-# cudaGetDriverEntryPointByVersion as a wrapper to avoid directly calling the 
-# driver API. This causes problems when linking with earlier versions of CUDA.
-# Setting this variable sidesteps the issue by calling the driver directly.
-target_compile_definitions(_C PRIVATE CUTLASS_ENABLE_DIRECT_CUDA_DRIVER_CALL=1)
-
 #
 # _moe_C extension
 #
 
 set(VLLM_MOE_EXT_SRC
-  "csrc/moe/torch_bindings.cpp"
+  "csrc/moe/moe_ops.cpp"
   "csrc/moe/topk_softmax_kernels.cu")
 
-if(VLLM_GPU_LANG STREQUAL "CUDA")
-  list(APPEND VLLM_MOE_EXT_SRC
-      "csrc/moe/marlin_kernels/marlin_moe_kernel.h"
-      "csrc/moe/marlin_kernels/marlin_moe_kernel_ku4b8.h"
-      "csrc/moe/marlin_kernels/marlin_moe_kernel_ku4b8.cu"
-      "csrc/moe/marlin_kernels/marlin_moe_kernel_ku8b128.h"
-      "csrc/moe/marlin_kernels/marlin_moe_kernel_ku8b128.cu"
-      "csrc/moe/marlin_moe_ops.cu")
-endif()
-
-message(STATUS "Enabling moe extension.")
 define_gpu_extension_target(
   _moe_C
   DESTINATION vllm
@@ -332,88 +202,101 @@ define_gpu_extension_target(
   SOURCES ${VLLM_MOE_EXT_SRC}
   COMPILE_FLAGS ${VLLM_GPU_FLAGS}
   ARCHITECTURES ${VLLM_GPU_ARCHES}
-  USE_SABI 3
   WITH_SOABI)
 
-if(VLLM_GPU_LANG STREQUAL "HIP")
-  #
-  # _rocm_C extension
-  #
-  set(VLLM_ROCM_EXT_SRC
-    "csrc/rocm/torch_bindings.cpp"
-    "csrc/rocm/attention.cu")
+#
+# _punica_C extension
+#
 
+set(VLLM_PUNICA_EXT_SRC
+  "csrc/punica/bgmv/bgmv_bf16_bf16_bf16.cu"
+  "csrc/punica/bgmv/bgmv_bf16_bf16_fp16.cu"
+  "csrc/punica/bgmv/bgmv_bf16_fp16_bf16.cu"
+  "csrc/punica/bgmv/bgmv_bf16_fp16_fp16.cu"
+  "csrc/punica/bgmv/bgmv_bf16_fp32_bf16.cu"
+  "csrc/punica/bgmv/bgmv_bf16_fp32_fp16.cu"
+  "csrc/punica/bgmv/bgmv_fp16_bf16_bf16.cu"
+  "csrc/punica/bgmv/bgmv_fp16_bf16_fp16.cu"
+  "csrc/punica/bgmv/bgmv_fp16_fp16_bf16.cu"
+  "csrc/punica/bgmv/bgmv_fp16_fp16_fp16.cu"
+  "csrc/punica/bgmv/bgmv_fp16_fp32_bf16.cu"
+  "csrc/punica/bgmv/bgmv_fp16_fp32_fp16.cu"
+  "csrc/punica/bgmv/bgmv_fp32_bf16_bf16.cu"
+  "csrc/punica/bgmv/bgmv_fp32_bf16_fp16.cu"
+  "csrc/punica/bgmv/bgmv_fp32_fp16_bf16.cu"
+  "csrc/punica/bgmv/bgmv_fp32_fp16_fp16.cu"
+  "csrc/punica/bgmv/bgmv_fp32_fp32_bf16.cu"
+  "csrc/punica/bgmv/bgmv_fp32_fp32_fp16.cu"
+  "csrc/punica/punica_ops.cc")
+
+#
+# Copy GPU compilation flags+update for punica
+#
+set(VLLM_PUNICA_GPU_FLAGS ${VLLM_GPU_FLAGS})
+list(REMOVE_ITEM VLLM_PUNICA_GPU_FLAGS
+  "-D__CUDA_NO_HALF_OPERATORS__"
+  "-D__CUDA_NO_HALF_CONVERSIONS__"
+  "-D__CUDA_NO_BFLOAT16_CONVERSIONS__"
+  "-D__CUDA_NO_HALF2_OPERATORS__")
+
+#
+# Filter out CUDA architectures < 8.0 for punica.
+#
+if (${VLLM_GPU_LANG} STREQUAL "CUDA")
+  set(VLLM_PUNICA_GPU_ARCHES)
+  foreach(ARCH ${VLLM_GPU_ARCHES})
+    string_to_ver(CODE_VER ${ARCH})
+    if (CODE_VER GREATER_EQUAL 8.0)
+      list(APPEND VLLM_PUNICA_GPU_ARCHES ${ARCH})
+    endif()
+  endforeach()
+  message(STATUS "Punica target arches: ${VLLM_PUNICA_GPU_ARCHES}")
+endif()
+
+if (VLLM_PUNICA_GPU_ARCHES)
   define_gpu_extension_target(
-    _rocm_C
+    _punica_C
     DESTINATION vllm
     LANGUAGE ${VLLM_GPU_LANG}
-    SOURCES ${VLLM_ROCM_EXT_SRC}
-    COMPILE_FLAGS ${VLLM_GPU_FLAGS}
-    ARCHITECTURES ${VLLM_GPU_ARCHES}
-    USE_SABI 3
+    SOURCES ${VLLM_PUNICA_EXT_SRC}
+    COMPILE_FLAGS ${VLLM_PUNICA_GPU_FLAGS}
+    ARCHITECTURES ${VLLM_PUNICA_GPU_ARCHES}
     WITH_SOABI)
-endif()
-
-# vllm-flash-attn currently only supported on CUDA
-if (NOT VLLM_TARGET_DEVICE STREQUAL "cuda")
-  return()
-endif ()
-
-#
-# Build vLLM flash attention from source
-#
-# IMPORTANT: This has to be the last thing we do, because vllm-flash-attn uses the same macros/functions as vLLM.
-# Because functions all belong to the global scope, vllm-flash-attn's functions overwrite vLLMs.
-# They should be identical but if they aren't, this is a massive footgun.
-#
-# The vllm-flash-attn install rules are nested under vllm to make sure the library gets installed in the correct place.
-# To only install vllm-flash-attn, use --component vllm_flash_attn_c.
-# If no component is specified, vllm-flash-attn is still installed.
-
-# If VLLM_FLASH_ATTN_SRC_DIR is set, vllm-flash-attn is installed from that directory instead of downloading.
-# This is to enable local development of vllm-flash-attn within vLLM.
-# It can be set as an environment variable or passed as a cmake argument.
-# The environment variable takes precedence.
-if (DEFINED ENV{VLLM_FLASH_ATTN_SRC_DIR})
-  set(VLLM_FLASH_ATTN_SRC_DIR $ENV{VLLM_FLASH_ATTN_SRC_DIR})
-endif()
-
-if(VLLM_FLASH_ATTN_SRC_DIR)
-  FetchContent_Declare(vllm-flash-attn SOURCE_DIR ${VLLM_FLASH_ATTN_SRC_DIR})
 else()
-  FetchContent_Declare(
-          vllm-flash-attn
-          GIT_REPOSITORY https://github.com/vllm-project/flash-attention.git
-          GIT_TAG 013f0c4fc47e6574060879d9734c1df8c5c273bd
-          GIT_PROGRESS TRUE
-  )
+  message(WARNING "Unable to create _punica_C target because none of the "
+    "requested architectures (${VLLM_GPU_ARCHES}) are supported, i.e. >= 8.0")
 endif()
 
-# Set the parent build flag so that the vllm-flash-attn library does not redo compile flag and arch initialization.
-set(VLLM_PARENT_BUILD ON)
+#
+# Add the `default` target which detects which extensions should be
+# built based on platform/architecture.  This is the same logic that
+# setup.py uses to select which extensions should be built and should
+# be kept in sync.
+#
+# The `default` target makes direct use of cmake easier since knowledge
+# of which extensions are supported has been factored in, e.g.
+#
+# mkdir build && cd build
+# cmake -G Ninja -DVLLM_PYTHON_EXECUTABLE=`which python3` -DCMAKE_LIBRARY_OUTPUT_DIRECTORY=../vllm ..
+# cmake --build . --target default
+#
+add_custom_target(default)
 
-# Ensure the vllm/vllm_flash_attn directory exists before installation
-install(CODE "file(MAKE_DIRECTORY \"\${CMAKE_INSTALL_PREFIX}/vllm/vllm_flash_attn\")" COMPONENT vllm_flash_attn_c)
+if(VLLM_GPU_LANG STREQUAL "CUDA" OR VLLM_GPU_LANG STREQUAL "HIP")
+  message(STATUS "Enabling C extension.")
+  add_dependencies(default _C)
+endif()
 
-# Make sure vllm-flash-attn install rules are nested under vllm/
-install(CODE "set(CMAKE_INSTALL_LOCAL_ONLY FALSE)" COMPONENT vllm_flash_attn_c)
-install(CODE "set(OLD_CMAKE_INSTALL_PREFIX \"\${CMAKE_INSTALL_PREFIX}\")" COMPONENT vllm_flash_attn_c)
-install(CODE "set(CMAKE_INSTALL_PREFIX \"\${CMAKE_INSTALL_PREFIX}/vllm/\")" COMPONENT vllm_flash_attn_c)
+if(VLLM_GPU_LANG STREQUAL "CUDA")
+  message(STATUS "Enabling moe extension.")
+  add_dependencies(default _moe_C)
 
-# Fetch the vllm-flash-attn library
-FetchContent_MakeAvailable(vllm-flash-attn)
-message(STATUS "vllm-flash-attn is available at ${vllm-flash-attn_SOURCE_DIR}")
-
-# Restore the install prefix
-install(CODE "set(CMAKE_INSTALL_PREFIX \"\${OLD_CMAKE_INSTALL_PREFIX}\")" COMPONENT vllm_flash_attn_c)
-install(CODE "set(CMAKE_INSTALL_LOCAL_ONLY TRUE)" COMPONENT vllm_flash_attn_c)
-
-# Copy over the vllm-flash-attn python files
-install(
-        DIRECTORY ${vllm-flash-attn_SOURCE_DIR}/vllm_flash_attn/
-        DESTINATION vllm/vllm_flash_attn
-        COMPONENT vllm_flash_attn_c
-        FILES_MATCHING PATTERN "*.py"
-)
-
-# Nothing after vllm-flash-attn, see comment about macros above
+  # Enable punica if -DVLLM_INSTALL_PUNICA_KERNELS=ON or
+  # VLLM_INSTALL_PUNICA_KERNELS is set in the environment and
+  # there are supported target arches.
+  if (VLLM_PUNICA_GPU_ARCHES AND
+      (ENV{VLLM_INSTALL_PUNICA_KERNELS} OR VLLM_INSTALL_PUNICA_KERNELS))
+    message(STATUS "Enabling punica extension.")
+    add_dependencies(default _punica_C)
+  endif()
+endif()

CODE_OF_CONDUCT.md

@@ -1,128 +0,0 @@
-
-# vLLM Code of Conduct
-
-## Our Pledge
-
-We as members, contributors, and leaders pledge to make participation in our
-community a harassment-free experience for everyone, regardless of age, body
-size, visible or invisible disability, ethnicity, sex characteristics, gender
-identity and expression, level of experience, education, socioeconomic status,
-nationality, personal appearance, race, caste, color, religion, or sexual
-identity and orientation.
-
-We pledge to act and interact in ways that contribute to an open, welcoming,
-diverse, inclusive, and healthy community.
-
-## Our Standards
-
-Examples of behavior that contributes to a positive environment for our
-community include:
-
-* Demonstrating empathy and kindness toward other people
-* Being respectful of differing opinions, viewpoints, and experiences
-* Giving and gracefully accepting constructive feedback
-* Accepting responsibility and apologizing to those affected by our mistakes,
-  and learning from the experience
-* Focusing on what is best not just for us as individuals, but for the overall
-  community
-
-Examples of unacceptable behavior include:
-
-* The use of sexualized language or imagery, and sexual attention or advances of
-  any kind
-* Trolling, insulting or derogatory comments, and personal or political attacks
-* Public or private harassment
-* Publishing others' private information, such as a physical or email address,
-  without their explicit permission
-* Other conduct which could reasonably be considered inappropriate in a
-  professional setting
-
-## Enforcement Responsibilities
-
-Community leaders are responsible for clarifying and enforcing our standards of
-acceptable behavior and will take appropriate and fair corrective action in
-response to any behavior that they deem inappropriate, threatening, offensive,
-or harmful.
-
-Community leaders have the right and responsibility to remove, edit, or reject
-comments, commits, code, wiki edits, issues, and other contributions that are
-not aligned to this Code of Conduct, and will communicate reasons for moderation
-decisions when appropriate.
-
-## Scope
-
-This Code of Conduct applies within all community spaces, and also applies when
-an individual is officially representing the community in public spaces.
-Examples of representing our community include using an official email address,
-posting via an official social media account, or acting as an appointed
-representative at an online or offline/IRL event.
-
-## Enforcement
-
-Instances of abusive, harassing, or otherwise unacceptable behavior may be
-reported to the community leaders responsible for enforcement in the #code-of-conduct
-channel in the [vLLM Discord](https://discord.com/invite/jz7wjKhh6g).
-All complaints will be reviewed and investigated promptly and fairly.
-
-All community leaders are obligated to respect the privacy and security of the
-reporter of any incident.
-
-## Enforcement Guidelines
-
-Community leaders will follow these Community Impact Guidelines in determining
-the consequences for any action they deem in violation of this Code of Conduct:
-
-### 1. Correction
-
-**Community Impact**: Use of inappropriate language or other behavior deemed
-unprofessional or unwelcome in the community.
-
-**Consequence**: A private, written warning from community leaders, providing
-clarity around the nature of the violation and an explanation of why the
-behavior was inappropriate. A public apology may be requested.
-
-### 2. Warning
-
-**Community Impact**: A violation through a single incident or series of
-actions.
-
-**Consequence**: A warning with consequences for continued behavior. No
-interaction with the people involved, including unsolicited interaction with
-those enforcing the Code of Conduct, for a specified period of time. This
-includes avoiding interactions in community spaces as well as external channels
-like social media. Violating these terms may lead to a temporary or permanent
-ban.
-
-### 3. Temporary Ban
-
-**Community Impact**: A serious violation of community standards, including
-sustained inappropriate behavior.
-
-**Consequence**: A temporary ban from any sort of interaction or public
-communication with the community for a specified period of time. No public or
-private interaction with the people involved, including unsolicited interaction
-with those enforcing the Code of Conduct, is allowed during this period.
-Violating these terms may lead to a permanent ban.
-
-### 4. Permanent Ban
-
-**Community Impact**: Demonstrating a pattern of violation of community
-standards, including sustained inappropriate behavior, harassment of an
-individual, or aggression toward or disparagement of classes of individuals.
-
-**Consequence**: A permanent ban from any sort of public interaction within the
-community.
-
-## Attribution
-
-This Code of Conduct is adapted from the [Contributor Covenant](https://www.contributor-covenant.org/),
-version 2.1, available at
-[v2.1](https://www.contributor-covenant.org/version/2/1/code_of_conduct.html).
-
-Community Impact Guidelines were inspired by
-[Mozilla's code of conduct enforcement ladder](https://github.com/mozilla/inclusion).
-
-For answers to common questions about this code of conduct, see the
-[Contributor Covenant FAQ](https://www.contributor-covenant.org/faq). Translations are available at
-[Contributor Covenant translations](https://www.contributor-covenant.org/translations).
-

Dockerfile

@@ -1,37 +1,18 @@
 # The vLLM Dockerfile is used to construct vLLM image that can be directly used
 # to run the OpenAI compatible server.
 
-# Please update any changes made here to
-# docs/source/dev/dockerfile/dockerfile.rst and
-# docs/source/assets/dev/dockerfile-stages-dependency.png
-
-ARG CUDA_VERSION=12.4.1
 #################### BASE BUILD IMAGE ####################
 # prepare basic build environment
-FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu20.04 AS base
-ARG CUDA_VERSION=12.4.1
-ARG PYTHON_VERSION=3.12
-ENV DEBIAN_FRONTEND=noninteractive
+FROM nvidia/cuda:12.1.0-devel-ubuntu22.04 AS dev
 
-# Install Python and other dependencies
-RUN echo 'tzdata tzdata/Areas select America' | debconf-set-selections \
-    && echo 'tzdata tzdata/Zones/America select Los_Angeles' | debconf-set-selections \
-    && apt-get update -y \
-    && apt-get install -y ccache software-properties-common git curl sudo \
-    && add-apt-repository ppa:deadsnakes/ppa \
-    && apt-get update -y \
-    && apt-get install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python${PYTHON_VERSION}-venv \
-    && update-alternatives --install /usr/bin/python3 python3 /usr/bin/python${PYTHON_VERSION} 1 \
-    && update-alternatives --set python3 /usr/bin/python${PYTHON_VERSION} \
-    && ln -sf /usr/bin/python${PYTHON_VERSION}-config /usr/bin/python3-config \
-    && curl -sS https://bootstrap.pypa.io/get-pip.py | python${PYTHON_VERSION} \
-    && python3 --version && python3 -m pip --version
+RUN apt-get update -y \
+    && apt-get install -y python3-pip git
 
 # Workaround for https://github.com/openai/triton/issues/2507 and
 # https://github.com/pytorch/pytorch/issues/107960 -- hopefully
 # this won't be needed for future versions of this docker image
 # or future versions of triton.
-RUN ldconfig /usr/local/cuda-$(echo $CUDA_VERSION | cut -d. -f1,2)/compat/
+RUN ldconfig /usr/local/cuda-12.1/compat/
 
 WORKDIR /workspace
 
@@ -39,8 +20,12 @@ WORKDIR /workspace
 COPY requirements-common.txt requirements-common.txt
 COPY requirements-cuda.txt requirements-cuda.txt
 RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install -r requirements-cuda.txt
+    pip install -r requirements-cuda.txt
 
+# install development dependencies
+COPY requirements-dev.txt requirements-dev.txt
+RUN --mount=type=cache,target=/root/.cache/pip \
+    pip install -r requirements-dev.txt
 
 # cuda arch list used by torch
 # can be useful for both `dev` and `test`
@@ -48,19 +33,19 @@ RUN --mount=type=cache,target=/root/.cache/pip \
 # see https://github.com/pytorch/pytorch/pull/123243
 ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0+PTX'
 ENV TORCH_CUDA_ARCH_LIST=${torch_cuda_arch_list}
-# Override the arch list for flash-attn to reduce the binary size
-ARG vllm_fa_cmake_gpu_arches='80-real;90-real'
-ENV VLLM_FA_CMAKE_GPU_ARCHES=${vllm_fa_cmake_gpu_arches}
 #################### BASE BUILD IMAGE ####################
 
+
 #################### WHEEL BUILD IMAGE ####################
-FROM base AS build
+FROM dev AS build
 
 # install build dependencies
 COPY requirements-build.txt requirements-build.txt
-
 RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install -r requirements-build.txt
+    pip install -r requirements-build.txt
+
+# install compiler cache to speed up compilation leveraging local or remote caching
+RUN apt-get update -y && apt-get install -y ccache
 
 # files and directories related to build wheels
 COPY csrc csrc
@@ -78,101 +63,61 @@ ENV MAX_JOBS=${max_jobs}
 # number of threads used by nvcc
 ARG nvcc_threads=8
 ENV NVCC_THREADS=$nvcc_threads
-
-ARG USE_SCCACHE
-ARG SCCACHE_BUCKET_NAME=vllm-build-sccache
-ARG SCCACHE_REGION_NAME=us-west-2
-ARG SCCACHE_S3_NO_CREDENTIALS=0
-# if USE_SCCACHE is set, use sccache to speed up compilation
-RUN --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=bind,source=.git,target=.git \
-    if [ "$USE_SCCACHE" = "1" ]; then \
-        echo "Installing sccache..." \
-        && curl -L -o sccache.tar.gz https://github.com/mozilla/sccache/releases/download/v0.8.1/sccache-v0.8.1-x86_64-unknown-linux-musl.tar.gz \
-        && tar -xzf sccache.tar.gz \
-        && sudo mv sccache-v0.8.1-x86_64-unknown-linux-musl/sccache /usr/bin/sccache \
-        && rm -rf sccache.tar.gz sccache-v0.8.1-x86_64-unknown-linux-musl \
-        && export SCCACHE_BUCKET=${SCCACHE_BUCKET_NAME} \
-        && export SCCACHE_REGION=${SCCACHE_REGION_NAME} \
-        && export SCCACHE_S3_NO_CREDENTIALS=${SCCACHE_S3_NO_CREDENTIALS} \
-        && export SCCACHE_IDLE_TIMEOUT=0 \
-        && export CMAKE_BUILD_TYPE=Release \
-        && sccache --show-stats \
-        && python3 setup.py bdist_wheel --dist-dir=dist --py-limited-api=cp38 \
-        && sccache --show-stats; \
-    fi
+# make sure punica kernels are built (for LoRA)
+ENV VLLM_INSTALL_PUNICA_KERNELS=1
 
 ENV CCACHE_DIR=/root/.cache/ccache
 RUN --mount=type=cache,target=/root/.cache/ccache \
     --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=bind,source=.git,target=.git  \
-    if [ "$USE_SCCACHE" != "1" ]; then \
-        python3 setup.py bdist_wheel --dist-dir=dist --py-limited-api=cp38; \
-    fi
+    python3 setup.py bdist_wheel --dist-dir=dist
 
-# Check the size of the wheel if RUN_WHEEL_CHECK is true
-COPY .buildkite/check-wheel-size.py check-wheel-size.py
-# Default max size of the wheel is 250MB
-ARG VLLM_MAX_SIZE_MB=250
-ENV VLLM_MAX_SIZE_MB=$VLLM_MAX_SIZE_MB
-ARG RUN_WHEEL_CHECK=true
-RUN if [ "$RUN_WHEEL_CHECK" = "true" ]; then \
-        python3 check-wheel-size.py dist; \
-    else \
-        echo "Skipping wheel size check."; \
-    fi
+# the `vllm_nccl` package must be installed from source distribution
+# pip is too smart to store a wheel in the cache, and other CI jobs
+# will directly use the wheel from the cache, which is not what we want.
+# we need to remove it manually
+RUN --mount=type=cache,target=/root/.cache/pip \
+    pip cache remove vllm_nccl*
 #################### EXTENSION Build IMAGE ####################
 
-#################### DEV IMAGE ####################
-FROM base as dev
+#################### FLASH_ATTENTION Build IMAGE ####################
+FROM dev as flash-attn-builder
+# max jobs used for build
+ARG max_jobs=2
+ENV MAX_JOBS=${max_jobs}
+# flash attention version
+ARG flash_attn_version=v2.5.6
+ENV FLASH_ATTN_VERSION=${flash_attn_version}
 
-COPY requirements-lint.txt requirements-lint.txt
-COPY requirements-test.txt requirements-test.txt
-COPY requirements-dev.txt requirements-dev.txt
-RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install -r requirements-dev.txt
+WORKDIR /usr/src/flash-attention-v2
+
+# Download the wheel or build it if a pre-compiled release doesn't exist
+RUN pip --verbose wheel flash-attn==${FLASH_ATTN_VERSION} \
+    --no-build-isolation --no-deps --no-cache-dir
+
+#################### FLASH_ATTENTION Build IMAGE ####################
 
-#################### DEV IMAGE ####################
 #################### vLLM installation IMAGE ####################
 # image with vLLM installed
-FROM nvidia/cuda:${CUDA_VERSION}-base-ubuntu20.04 AS vllm-base
-ARG CUDA_VERSION=12.4.1
-ARG PYTHON_VERSION=3.12
+FROM nvidia/cuda:12.1.0-base-ubuntu22.04 AS vllm-base
 WORKDIR /vllm-workspace
-ENV DEBIAN_FRONTEND=noninteractive
 
-RUN PYTHON_VERSION_STR=$(echo ${PYTHON_VERSION} | sed 's/\.//g') && \
-    echo "export PYTHON_VERSION_STR=${PYTHON_VERSION_STR}" >> /etc/environment
-
-# Install Python and other dependencies
-RUN echo 'tzdata tzdata/Areas select America' | debconf-set-selections \
-    && echo 'tzdata tzdata/Zones/America select Los_Angeles' | debconf-set-selections \
-    && apt-get update -y \
-    && apt-get install -y ccache software-properties-common git curl sudo vim python3-pip \
-    && apt-get install -y ffmpeg libsm6 libxext6 libgl1 \
-    && add-apt-repository ppa:deadsnakes/ppa \
-    && apt-get update -y \
-    && apt-get install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python${PYTHON_VERSION}-venv libibverbs-dev \
-    && update-alternatives --install /usr/bin/python3 python3 /usr/bin/python${PYTHON_VERSION} 1 \
-    && update-alternatives --set python3 /usr/bin/python${PYTHON_VERSION} \
-    && ln -sf /usr/bin/python${PYTHON_VERSION}-config /usr/bin/python3-config \
-    && curl -sS https://bootstrap.pypa.io/get-pip.py | python${PYTHON_VERSION} \
-    && python3 --version && python3 -m pip --version
+RUN apt-get update -y \
+    && apt-get install -y python3-pip git vim
 
 # Workaround for https://github.com/openai/triton/issues/2507 and
 # https://github.com/pytorch/pytorch/issues/107960 -- hopefully
 # this won't be needed for future versions of this docker image
 # or future versions of triton.
-RUN ldconfig /usr/local/cuda-$(echo $CUDA_VERSION | cut -d. -f1,2)/compat/
+RUN ldconfig /usr/local/cuda-12.1/compat/
 
 # install vllm wheel first, so that torch etc will be installed
 RUN --mount=type=bind,from=build,src=/workspace/dist,target=/vllm-workspace/dist \
     --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install dist/*.whl --verbose
+    pip install dist/*.whl --verbose
 
-RUN --mount=type=cache,target=/root/.cache/pip \
-    . /etc/environment && \
-    python3 -m pip install https://github.com/flashinfer-ai/flashinfer/releases/download/v0.1.6/flashinfer-0.1.6+cu121torch2.4-cp${PYTHON_VERSION_STR}-cp${PYTHON_VERSION_STR}-linux_x86_64.whl
+RUN --mount=type=bind,from=flash-attn-builder,src=/usr/src/flash-attention-v2,target=/usr/src/flash-attention-v2 \
+    --mount=type=cache,target=/root/.cache/pip \
+    pip install /usr/src/flash-attention-v2/*.whl --no-cache-dir
 #################### vLLM installation IMAGE ####################
 
 
@@ -185,7 +130,7 @@ ADD . /vllm-workspace/
 
 # install development dependencies (for testing)
 RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install -r requirements-dev.txt
+    pip install -r requirements-dev.txt
 
 # doc requires source code
 # we hide them inside `test_docs/` , so that this source code
@@ -202,7 +147,7 @@ FROM vllm-base AS vllm-openai
 
 # install additional dependencies for openai api server
 RUN --mount=type=cache,target=/root/.cache/pip \
-    pip install accelerate hf_transfer 'modelscope!=1.15.0' bitsandbytes>=0.44.0 timm==0.9.10
+    pip install accelerate hf_transfer modelscope
 
 ENV VLLM_USAGE_SOURCE production-docker-image
 

Dockerfile.cpu

@@ -1,74 +1,20 @@
 # This vLLM Dockerfile is used to construct image that can build and run vLLM on x86 CPU platform.
 
-FROM ubuntu:22.04 AS cpu-test-1
+FROM ubuntu:22.04
 
-ENV CCACHE_DIR=/root/.cache/ccache
-
-ENV CMAKE_CXX_COMPILER_LAUNCHER=ccache
-
-RUN --mount=type=cache,target=/var/cache/apt \
-    apt-get update -y \
-    && apt-get install -y curl ccache git wget vim numactl gcc-12 g++-12 python3 python3-pip libtcmalloc-minimal4 libnuma-dev \
-    && apt-get install -y ffmpeg libsm6 libxext6 libgl1 \
+RUN apt-get update  -y \
+    && apt-get install -y git wget vim numactl gcc-12 g++-12 python3 python3-pip \
     && update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-12 10 --slave /usr/bin/g++ g++ /usr/bin/g++-12
 
-# https://intel.github.io/intel-extension-for-pytorch/cpu/latest/tutorials/performance_tuning/tuning_guide.html
-# intel-openmp provides additional performance improvement vs. openmp
-# tcmalloc provides better memory allocation efficiency, e.g, holding memory in caches to speed up access of commonly-used objects.
-RUN --mount=type=cache,target=/root/.cache/pip \
-    pip install intel-openmp
+RUN pip install --upgrade pip \
+    && pip install wheel packaging ninja setuptools>=49.4.0 numpy
 
-ENV LD_PRELOAD="/usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4:/usr/local/lib/libiomp5.so"
-
-RUN echo 'ulimit -c 0' >> ~/.bashrc
-
-RUN pip install https://intel-extension-for-pytorch.s3.amazonaws.com/ipex_dev/cpu/intel_extension_for_pytorch-2.4.0%2Bgitfbaa4bc-cp310-cp310-linux_x86_64.whl
-
-WORKDIR /workspace
-
-ENV PIP_EXTRA_INDEX_URL=https://download.pytorch.org/whl/cpu
-RUN --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=bind,src=requirements-build.txt,target=requirements-build.txt \
-    pip install --upgrade pip && \
-    pip install -r requirements-build.txt
-
-# install oneDNN
-RUN git clone -b rls-v3.5 https://github.com/oneapi-src/oneDNN.git
-
-RUN --mount=type=cache,target=/root/.cache/ccache \
-    cmake -B ./oneDNN/build -S ./oneDNN -G Ninja -DONEDNN_LIBRARY_TYPE=STATIC \ 
-    -DONEDNN_BUILD_DOC=OFF \ 
-    -DONEDNN_BUILD_EXAMPLES=OFF \ 
-    -DONEDNN_BUILD_TESTS=OFF \ 
-    -DONEDNN_BUILD_GRAPH=OFF \ 
-    -DONEDNN_ENABLE_WORKLOAD=INFERENCE \ 
-    -DONEDNN_ENABLE_PRIMITIVE=MATMUL && \
-    cmake --build ./oneDNN/build --target install --config Release
-
-FROM cpu-test-1 AS build
+COPY ./ /workspace/vllm
 
 WORKDIR /workspace/vllm
 
-RUN --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=bind,src=requirements-common.txt,target=requirements-common.txt \
-    --mount=type=bind,src=requirements-cpu.txt,target=requirements-cpu.txt \
-    pip install -v -r requirements-cpu.txt
+RUN pip install -v -r requirements-cpu.txt --extra-index-url https://download.pytorch.org/whl/cpu
 
-COPY ./ ./
+RUN VLLM_TARGET_DEVICE=cpu python3 setup.py install
 
-# Support for building with non-AVX512 vLLM: docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" ...
-ARG VLLM_CPU_DISABLE_AVX512
-ENV VLLM_CPU_DISABLE_AVX512=${VLLM_CPU_DISABLE_AVX512}
-
-RUN --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=cache,target=/root/.cache/ccache \
-    --mount=type=bind,source=.git,target=.git \
-    VLLM_TARGET_DEVICE=cpu python3 setup.py bdist_wheel && \
-    pip install dist/*.whl && \
-    rm -rf dist
-
-WORKDIR /workspace/
-
-RUN ln -s /workspace/vllm/tests && ln -s /workspace/vllm/examples && ln -s /workspace/vllm/benchmarks
-
-ENTRYPOINT ["python3", "-m", "vllm.entrypoints.openai.api_server"]
+CMD ["/bin/bash"]

Dockerfile.neuron

@@ -1,41 +0,0 @@
-# default base image
-ARG BASE_IMAGE="public.ecr.aws/neuron/pytorch-inference-neuronx:2.1.2-neuronx-py310-sdk2.20.0-ubuntu20.04"
-
-FROM $BASE_IMAGE
-
-RUN echo "Base image is $BASE_IMAGE"
-
-# Install some basic utilities
-RUN apt-get update && \
-    apt-get install -y \
-        git \
-        python3 \
-        python3-pip \
-        ffmpeg libsm6 libxext6 libgl1
-
-### Mount Point ###
-# When launching the container, mount the code directory to /app
-ARG APP_MOUNT=/app
-VOLUME [ ${APP_MOUNT} ]
-WORKDIR ${APP_MOUNT}
-
-RUN python3 -m pip install --upgrade pip
-RUN python3 -m pip install --no-cache-dir fastapi ninja tokenizers pandas
-RUN python3 -m pip install sentencepiece transformers==4.36.2 -U
-RUN python3 -m pip install transformers-neuronx --extra-index-url=https://pip.repos.neuron.amazonaws.com -U
-RUN python3 -m pip install --pre neuronx-cc==2.15.* --extra-index-url=https://pip.repos.neuron.amazonaws.com -U
-
-COPY . /app/vllm
-
-RUN cd /app/vllm \
-    && python3 -m pip install -U \
-        cmake>=3.26 ninja packaging setuptools-scm>=8 wheel jinja2 \
-        -r requirements-neuron.txt
-
-ENV VLLM_TARGET_DEVICE neuron
-RUN --mount=type=bind,source=.git,target=.git \
-    cd /app/vllm \
-    && pip install --no-build-isolation -v -e . \
-    && cd ..
-
-CMD ["/bin/bash"]

Dockerfile.openvino

@@ -1,31 +0,0 @@
-# The vLLM Dockerfile is used to construct vLLM image that can be directly used
-# to run the OpenAI compatible server.
-
-FROM ubuntu:22.04 AS dev
-
-RUN apt-get update -y && \
-    apt-get install -y \
-        git python3-pip \
-        ffmpeg libsm6 libxext6 libgl1
-WORKDIR /workspace
-
-# copy requirements
-COPY requirements-build.txt /workspace/vllm/
-COPY requirements-common.txt /workspace/vllm/
-COPY requirements-openvino.txt /workspace/vllm/
-
-COPY vllm/ /workspace/vllm/vllm
-COPY csrc/core /workspace/vllm/csrc/core
-COPY cmake/utils.cmake /workspace/vllm/cmake/
-COPY CMakeLists.txt /workspace/vllm/
-COPY setup.py /workspace/vllm/
-
-# install build requirements
-RUN PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu" python3 -m pip install -r /workspace/vllm/requirements-build.txt
-# build vLLM with OpenVINO backend
-RUN PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu" VLLM_TARGET_DEVICE="openvino" python3 -m pip install /workspace/vllm/
-
-COPY examples/ /workspace/vllm/examples
-COPY benchmarks/ /workspace/vllm/benchmarks
-
-CMD ["/bin/bash"]

Dockerfile.ppc64le

@@ -1,33 +0,0 @@
-FROM mambaorg/micromamba
-ARG MAMBA_DOCKERFILE_ACTIVATE=1
-USER root
-
-ENV PATH="/usr/local/cargo/bin:$PATH:/opt/conda/bin/"
-
-RUN apt-get update -y && apt-get install -y git wget curl vim libnuma-dev libsndfile-dev libprotobuf-dev build-essential ffmpeg libsm6 libxext6 libgl1 
-
-# Some packages in requirements-cpu are installed here
-# IBM provides optimized packages for ppc64le processors in the open-ce project for mamba
-# Currently these may not be available for venv or pip directly
-RUN micromamba install -y -n base -c https://ftp.osuosl.org/pub/open-ce/1.11.0-p10/ -c defaults python=3.10 torchvision-cpu=0.16.2 rust && micromamba clean --all --yes
-
-COPY ./ /workspace/vllm
-
-WORKDIR /workspace/vllm
-
-# These packages will be in rocketce eventually
-RUN --mount=type=cache,target=/root/.cache/pip  \
-    pip install -v --prefer-binary --extra-index-url https://repo.fury.io/mgiessing \
-        cmake>=3.26 ninja packaging setuptools-scm>=8 wheel jinja2 \
-        torch==2.3.1 \
-        -r requirements-cpu.txt \
-        xformers uvloop==0.20.0
-
-RUN --mount=type=bind,source=.git,target=.git \
-    VLLM_TARGET_DEVICE=cpu python3 setup.py install
-
-WORKDIR /workspace/
-
-RUN ln -s /workspace/vllm/tests && ln -s /workspace/vllm/examples && ln -s /workspace/vllm/benchmarks
-
-ENTRYPOINT ["python3", "-m", "vllm.entrypoints.openai.api_server"]

Dockerfile.rocm

@@ -1,27 +1,35 @@
-# Default ROCm 6.2 base image
-ARG BASE_IMAGE="rocm/pytorch:rocm6.2_ubuntu20.04_py3.9_pytorch_release_2.3.0"
+# default base image
+ARG BASE_IMAGE="rocm/pytorch:rocm6.0_ubuntu20.04_py3.9_pytorch_2.1.1"
+
+FROM $BASE_IMAGE
+
+ARG BASE_IMAGE="rocm/pytorch:rocm6.0_ubuntu20.04_py3.9_pytorch_2.1.1"
+
+RUN echo "Base image is $BASE_IMAGE"
+
+# BASE_IMAGE for ROCm_5.7: "rocm/pytorch:rocm5.7_ubuntu22.04_py3.10_pytorch_2.0.1"
+# BASE_IMAGE for ROCm_6.0: "rocm/pytorch:rocm6.0_ubuntu20.04_py3.9_pytorch_2.1.1"
 
-# Default ROCm ARCHes to build vLLM for.
-ARG PYTORCH_ROCM_ARCH="gfx908;gfx90a;gfx942;gfx1100"
 
-# Whether to install CK-based flash-attention
-# If 0, will not install flash-attention
-ARG BUILD_FA="1"
 ARG FA_GFX_ARCHS="gfx90a;gfx942"
-ARG FA_BRANCH="3cea2fb"
+RUN echo "FA_GFX_ARCHS is $FA_GFX_ARCHS"
 
-# Whether to build triton on rocm
+ARG FA_BRANCH="3d2b6f5"
+RUN echo "FA_BRANCH is $FA_BRANCH"
+
+# whether to build flash-attention
+# if 0, will not build flash attention
+# this is useful for gfx target where flash-attention is not supported
+# In that case, we need to use the python reference attention implementation in vllm
+ARG BUILD_FA="1"
+
+# whether to build triton on rocm
 ARG BUILD_TRITON="1"
-ARG TRITON_BRANCH="e192dba"
-
-### Base image build stage
-FROM $BASE_IMAGE AS base
-
-# Import arg(s) defined before this build stage
-ARG PYTORCH_ROCM_ARCH
 
 # Install some basic utilities
 RUN apt-get update && apt-get install python3 python3-pip -y
+
+# Install some basic utilities
 RUN apt-get update && apt-get install -y \
     curl \
     ca-certificates \
@@ -32,135 +40,70 @@ RUN apt-get update && apt-get install -y \
     build-essential \
     wget \
     unzip \
+    nvidia-cuda-toolkit \
     tmux \
-    ccache \
  && rm -rf /var/lib/apt/lists/*
 
-# When launching the container, mount the code directory to /vllm-workspace
-ARG APP_MOUNT=/vllm-workspace
+### Mount Point ###
+# When launching the container, mount the code directory to /app
+ARG APP_MOUNT=/app
+VOLUME [ ${APP_MOUNT} ]
 WORKDIR ${APP_MOUNT}
 
 RUN python3 -m pip install --upgrade pip
-# Remove sccache so it doesn't interfere with ccache
-# TODO: implement sccache support across components
-RUN apt-get purge -y sccache; python3 -m pip uninstall -y sccache; rm -f "$(which sccache)"
-
-# Install torch == 2.6.0 on ROCm
-RUN --mount=type=cache,target=/root/.cache/pip \
-    case "$(ls /opt | grep -Po 'rocm-[0-9]\.[0-9]')" in \
-        *"rocm-6.2"*) \
-            python3 -m pip uninstall -y torch torchvision \
-            && python3 -m pip install --pre \
-                torch==2.6.0.dev20240918 \
-                setuptools-scm>=8 \
-                torchvision==0.20.0.dev20240918 \
-                --extra-index-url https://download.pytorch.org/whl/nightly/rocm6.2;; \
-        *) ;; esac
+RUN python3 -m pip install --no-cache-dir fastapi ninja tokenizers pandas
 
 ENV LLVM_SYMBOLIZER_PATH=/opt/rocm/llvm/bin/llvm-symbolizer
 ENV PATH=$PATH:/opt/rocm/bin:/libtorch/bin:
 ENV LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/rocm/lib/:/libtorch/lib:
 ENV CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/libtorch/include:/libtorch/include/torch/csrc/api/include/:/opt/rocm/include/:
 
-ENV PYTORCH_ROCM_ARCH=${PYTORCH_ROCM_ARCH}
-ENV CCACHE_DIR=/root/.cache/ccache
-
-
-### AMD-SMI build stage
-FROM base AS build_amdsmi
-# Build amdsmi wheel always
-RUN cd /opt/rocm/share/amd_smi \
-    && python3 -m pip wheel . --wheel-dir=/install
-
-
-### Flash-Attention wheel build stage
-FROM base AS build_fa
-ARG BUILD_FA
-ARG FA_GFX_ARCHS
-ARG FA_BRANCH
-# Build ROCm flash-attention wheel if `BUILD_FA = 1`
-RUN --mount=type=cache,target=${CCACHE_DIR} \
-    if [ "$BUILD_FA" = "1" ]; then \
-        mkdir -p libs \
-        && cd libs \
-        && git clone https://github.com/ROCm/flash-attention.git \
-        && cd flash-attention \
-        && git checkout "${FA_BRANCH}" \
-        && git submodule update --init \
-        && GPU_ARCHS="${FA_GFX_ARCHS}" python3 setup.py bdist_wheel --dist-dir=/install; \
-    # Create an empty directory otherwise as later build stages expect one
-    else mkdir -p /install; \
+# Install ROCm flash-attention
+RUN if [ "$BUILD_FA" = "1" ]; then \
+    mkdir libs \
+    && cd libs \
+    && git clone https://github.com/ROCm/flash-attention.git \
+    && cd flash-attention \
+    && git checkout ${FA_BRANCH} \
+    && git submodule update --init \
+    && export GPU_ARCHS=${FA_GFX_ARCHS} \
+    && if [ "$BASE_IMAGE" = "rocm/pytorch:rocm5.7_ubuntu22.04_py3.10_pytorch_2.0.1" ]; then \
+        patch /opt/conda/envs/py_3.10/lib/python3.10/site-packages/torch/utils/hipify/hipify_python.py hipify_patch.patch; fi \
+    && python3 setup.py install \
+    && cd ..; \
     fi
 
+# Error related to odd state for numpy 1.20.3 where there is no METADATA etc, but an extra LICENSES_bundled.txt.
+# Manually removed it so that later steps of numpy upgrade can continue
+RUN if [ "$BASE_IMAGE" = "rocm/pytorch:rocm6.0_ubuntu20.04_py3.9_pytorch_2.1.1" ]; then \
+    rm -rf /opt/conda/envs/py_3.9/lib/python3.9/site-packages/numpy-1.20.3.dist-info/; fi
 
-### Triton wheel build stage
-FROM base AS build_triton
-ARG BUILD_TRITON
-ARG TRITON_BRANCH
-# Build triton wheel if `BUILD_TRITON = 1`
-RUN --mount=type=cache,target=${CCACHE_DIR} \
-    if [ "$BUILD_TRITON" = "1" ]; then \
+# build triton
+RUN if [ "$BUILD_TRITON" = "1" ]; then \
     mkdir -p libs \
     && cd libs \
-    && python3 -m pip install ninja cmake wheel pybind11 \
-    && git clone https://github.com/OpenAI/triton.git \
-    && cd triton \
-    && git checkout "${TRITON_BRANCH}" \
-    && cd python \
-    && python3 setup.py bdist_wheel --dist-dir=/install; \
-    # Create an empty directory otherwise as later build stages expect one
-    else mkdir -p /install; \
+    && pip uninstall -y triton \
+    && git clone https://github.com/ROCm/triton.git \
+    && cd triton/python \
+    && pip3 install . \
+    && cd ../..; \
     fi
 
+COPY ./ /app/vllm
 
-### Final vLLM build stage
-FROM base AS final
-# Import the vLLM development directory from the build context
-COPY . .
+RUN python3 -m pip install --upgrade pip numba
+RUN python3 -m pip install xformers==0.0.23 --no-deps
 
-# Package upgrades for useful functionality or to avoid dependency issues
-RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install --upgrade numba scipy huggingface-hub[cli] pytest-shard
+RUN cd /app \
+    && cd vllm \
+    && pip install -U -r requirements-rocm.txt \
+    && if [ "$BUILD_FA" = "1" ]; then \
+       bash patch_xformers.rocm.sh; fi \
+    && patch /opt/rocm/include/hip/amd_detail/amd_hip_bf16.h /app/vllm/rocm_patch/rocm_bf16.patch \
+    && python3 setup.py install \
+    && cd ..
 
-
-# Workaround for ray >= 2.10.0
-ENV RAY_EXPERIMENTAL_NOSET_ROCR_VISIBLE_DEVICES=1
-# Silences the HF Tokenizers warning
-ENV TOKENIZERS_PARALLELISM=false
-
-RUN --mount=type=cache,target=${CCACHE_DIR} \
-    --mount=type=bind,source=.git,target=.git \
-    --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install -Ur requirements-rocm.txt \
-    && python3 setup.py clean --all \
-    && python3 setup.py develop
-
-# Copy amdsmi wheel into final image
-RUN --mount=type=bind,from=build_amdsmi,src=/install,target=/install \
-    mkdir -p libs \
-    && cp /install/*.whl libs \
-    # Preemptively uninstall to avoid same-version no-installs
-    && python3 -m pip uninstall -y amdsmi;
-
-# Copy triton wheel(s) into final image if they were built
-RUN --mount=type=bind,from=build_triton,src=/install,target=/install \
-    mkdir -p libs \
-    && if ls /install/*.whl; then \
-        cp /install/*.whl libs \
-        # Preemptively uninstall to avoid same-version no-installs
-        && python3 -m pip uninstall -y triton; fi
-
-# Copy flash-attn wheel(s) into final image if they were built
-RUN --mount=type=bind,from=build_fa,src=/install,target=/install \
-    mkdir -p libs \
-    && if ls /install/*.whl; then \
-        cp /install/*.whl libs \
-        # Preemptively uninstall to avoid same-version no-installs
-        && python3 -m pip uninstall -y flash-attn; fi
-
-# Install wheels that were built to the final image
-RUN --mount=type=cache,target=/root/.cache/pip \
-    if ls libs/*.whl; then \
-    python3 -m pip install libs/*.whl; fi
+RUN python3 -m pip install --upgrade pip
+RUN python3 -m pip install --no-cache-dir ray[all]==2.9.3
 
 CMD ["/bin/bash"]

Dockerfile.tpu

@@ -1,29 +0,0 @@
-ARG NIGHTLY_DATE="20240828"
-ARG BASE_IMAGE="us-central1-docker.pkg.dev/tpu-pytorch-releases/docker/xla:nightly_3.10_tpuvm_$NIGHTLY_DATE"
-
-FROM $BASE_IMAGE
-WORKDIR /workspace
-
-# Install some basic utilities
-RUN apt-get update && apt-get install -y \
-    git \
-    ffmpeg libsm6 libxext6 libgl1
-
-# Install the TPU and Pallas dependencies.
-RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install torch_xla[tpu] -f https://storage.googleapis.com/libtpu-releases/index.html
-RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install torch_xla[pallas] -f https://storage.googleapis.com/jax-releases/jax_nightly_releases.html -f https://storage.googleapis.com/jax-releases/jaxlib_nightly_releases.html
-
-# Build vLLM.
-COPY . /workspace/vllm
-ENV VLLM_TARGET_DEVICE="tpu"
-RUN --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=bind,source=.git,target=.git \
-     cd /workspace/vllm && \
-    python3 -m pip install \
-        cmake>=3.26 ninja packaging setuptools-scm>=8 wheel jinja2 \
-        -r requirements-tpu.txt
-RUN cd /workspace/vllm && python3 setup.py develop
-
-CMD ["/bin/bash"]

Dockerfile.xpu

@@ -1,26 +0,0 @@
-FROM intel/oneapi-basekit:2024.2.1-0-devel-ubuntu22.04
-
-RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | tee /usr/share/keyrings/intel-oneapi-archive-keyring.gpg > /dev/null && \
-    echo "deb [signed-by=/usr/share/keyrings/intel-oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main " | tee /etc/apt/sources.list.d/oneAPI.list && \
-    chmod 644 /usr/share/keyrings/intel-oneapi-archive-keyring.gpg && \
-    wget -O- https://repositories.intel.com/graphics/intel-graphics.key | gpg --dearmor | tee /usr/share/keyrings/intel-graphics.gpg > /dev/null && \
-    echo "deb [arch=amd64,i386 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://repositories.intel.com/graphics/ubuntu jammy arc" | tee /etc/apt/sources.list.d/intel.gpu.jammy.list && \
-    chmod 644 /usr/share/keyrings/intel-graphics.gpg
-
-RUN apt-get update  -y && \
-    apt-get install -y curl libicu70 lsb-release git wget vim numactl python3 python3-pip ffmpeg libsm6 libxext6 libgl1
-
-COPY ./ /workspace/vllm
-
-WORKDIR /workspace/vllm
-
-RUN --mount=type=cache,target=/root/.cache/pip \
-    pip install -v --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/ \
-        cmake>=3.26 ninja packaging setuptools-scm>=8 wheel jinja2 \
-        -r requirements-xpu.txt
-
-RUN --mount=type=cache,target=/root/.cache/pip \
-    --mount=type=bind,source=.git,target=.git \
-    VLLM_TARGET_DEVICE=xpu python3 setup.py install
-
-CMD ["/bin/bash"]

MANIFEST.in

@@ -1,9 +1,6 @@
 include LICENSE
 include requirements-common.txt
 include requirements-cuda.txt
-include requirements-rocm.txt
-include requirements-neuron.txt
-include requirements-cpu.txt
 include CMakeLists.txt
 
 recursive-include cmake *

README.md

@@ -10,30 +10,21 @@ Easy, fast, and cheap LLM serving for everyone
 </h3>
 
 <p align="center">
-| <a href="https://docs.vllm.ai"><b>Documentation</b></a> | <a href="https://vllm.ai"><b>Blog</b></a> | <a href="https://arxiv.org/abs/2309.06180"><b>Paper</b></a> | <a href="https://discord.gg/jz7wjKhh6g"><b>Discord</b></a> | <a href="https://x.com/vllm_project"><b>Twitter/X</b></a> |
+| <a href="https://docs.vllm.ai"><b>Documentation</b></a> | <a href="https://vllm.ai"><b>Blog</b></a> | <a href="https://arxiv.org/abs/2309.06180"><b>Paper</b></a> | <a href="https://discord.gg/jz7wjKhh6g"><b>Discord</b></a> |
 
 </p>
 
-
----
-
-**vLLM, AMD, Anyscale Meet & Greet at [Ray Summit 2024](http://raysummit.anyscale.com) (Monday, Sept 30th, 5-7pm PT) at Marriott Marquis San Francisco**
-
-We are excited to announce our special vLLM event in collaboration with AMD and Anyscale.
-Join us to learn more about recent advancements of vLLM on MI300X.
-Register [here](https://lu.ma/db5ld9n5) and be a part of the event!
-
----
-
 *Latest News* 🔥
-- [2024/09] We hosted [the sixth vLLM meetup](https://lu.ma/87q3nvnh) with NVIDIA! Please find the meetup slides [here](https://docs.google.com/presentation/d/1wrLGwytQfaOTd5wCGSPNhoaW3nq0E-9wqyP7ny93xRs/edit?usp=sharing).
-- [2024/07] We hosted [the fifth vLLM meetup](https://lu.ma/lp0gyjqr) with AWS! Please find the meetup slides [here](https://docs.google.com/presentation/d/1RgUD8aCfcHocghoP3zmXzck9vX3RCI9yfUAB2Bbcl4Y/edit?usp=sharing).
-- [2024/07] In partnership with Meta, vLLM officially supports Llama 3.1 with FP8 quantization and pipeline parallelism! Please check out our blog post [here](https://blog.vllm.ai/2024/07/23/llama31.html).
-- [2024/06] We hosted [the fourth vLLM meetup](https://lu.ma/agivllm) with Cloudflare and BentoML! Please find the meetup slides [here](https://docs.google.com/presentation/d/1iJ8o7V2bQEi0BFEljLTwc5G1S10_Rhv3beed5oB0NJ4/edit?usp=sharing).
 - [2024/04] We hosted [the third vLLM meetup](https://robloxandvllmmeetup2024.splashthat.com/) with Roblox! Please find the meetup slides [here](https://docs.google.com/presentation/d/1A--47JAK4BJ39t954HyTkvtfwn0fkqtsL8NGFuslReM/edit?usp=sharing).
-- [2024/01] We hosted [the second vLLM meetup](https://lu.ma/ygxbpzhl) with IBM! Please find the meetup slides [here](https://docs.google.com/presentation/d/12mI2sKABnUw5RBWXDYY-HtHth4iMSNcEoQ10jDQbxgA/edit?usp=sharing).
-- [2023/10] We hosted [the first vLLM meetup](https://lu.ma/first-vllm-meetup) with a16z! Please find the meetup slides [here](https://docs.google.com/presentation/d/1QL-XPFXiFpDBh86DbEegFXBXFXjix4v032GhShbKf3s/edit?usp=sharing).
+- [2024/01] We hosted [the second vLLM meetup](https://lu.ma/ygxbpzhl) in SF! Please find the meetup slides [here](https://docs.google.com/presentation/d/12mI2sKABnUw5RBWXDYY-HtHth4iMSNcEoQ10jDQbxgA/edit?usp=sharing).
+- [2024/01] Added ROCm 6.0 support to vLLM.
+- [2023/12] Added ROCm 5.7 support to vLLM.
+- [2023/10] We hosted [the first vLLM meetup](https://lu.ma/first-vllm-meetup) in SF! Please find the meetup slides [here](https://docs.google.com/presentation/d/1QL-XPFXiFpDBh86DbEegFXBXFXjix4v032GhShbKf3s/edit?usp=sharing).
+- [2023/09] We created our [Discord server](https://discord.gg/jz7wjKhh6g)! Join us to discuss vLLM and LLM serving! We will also post the latest announcements and updates there.
+- [2023/09] We released our [PagedAttention paper](https://arxiv.org/abs/2309.06180) on arXiv!
 - [2023/08] We would like to express our sincere gratitude to [Andreessen Horowitz](https://a16z.com/2023/08/30/supporting-the-open-source-ai-community/) (a16z) for providing a generous grant to support the open-source development and research of vLLM.
+- [2023/07] Added support for LLaMA-2! You can run and serve 7B/13B/70B LLaMA-2s on vLLM with a single command!
+- [2023/06] Serving vLLM On any Cloud with SkyPilot. Check out a 1-click [example](https://github.com/skypilot-org/skypilot/blob/master/llm/vllm) to start the vLLM demo, and the [blog post](https://blog.skypilot.co/serving-llm-24x-faster-on-the-cloud-with-vllm-and-skypilot/) for the story behind vLLM development on the clouds.
 - [2023/06] We officially released vLLM! FastChat-vLLM integration has powered [LMSYS Vicuna and Chatbot Arena](https://chat.lmsys.org) since mid-April. Check out our [blog post](https://vllm.ai).
 
 ---
@@ -46,41 +37,64 @@ vLLM is fast with:
 - Efficient management of attention key and value memory with **PagedAttention**
 - Continuous batching of incoming requests
 - Fast model execution with CUDA/HIP graph
-- Quantizations: [GPTQ](https://arxiv.org/abs/2210.17323), [AWQ](https://arxiv.org/abs/2306.00978), INT4, INT8, and FP8.
-- Optimized CUDA kernels, including integration with FlashAttention and FlashInfer.
-- Speculative decoding
-- Chunked prefill
-
-**Performance benchmark**: We include a [performance benchmark](https://buildkite.com/vllm/performance-benchmark/builds/4068) that compares the performance of vLLM against other LLM serving engines ([TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [text-generation-inference](https://github.com/huggingface/text-generation-inference) and [lmdeploy](https://github.com/InternLM/lmdeploy)).
+- Quantization: [GPTQ](https://arxiv.org/abs/2210.17323), [AWQ](https://arxiv.org/abs/2306.00978), [SqueezeLLM](https://arxiv.org/abs/2306.07629), FP8 KV Cache
+- Optimized CUDA kernels
 
 vLLM is flexible and easy to use with:
 
 - Seamless integration with popular Hugging Face models
 - High-throughput serving with various decoding algorithms, including *parallel sampling*, *beam search*, and more
-- Tensor parallelism and pipeline parallelism support for distributed inference
+- Tensor parallelism support for distributed inference
 - Streaming outputs
 - OpenAI-compatible API server
-- Support NVIDIA GPUs, AMD CPUs and GPUs, Intel CPUs and GPUs, PowerPC CPUs, TPU, and AWS Neuron.
-- Prefix caching support
-- Multi-lora support
+- Support NVIDIA GPUs and AMD GPUs
+- (Experimental) Prefix caching support
+- (Experimental) Multi-lora support
 
-vLLM seamlessly supports most popular open-source models on HuggingFace, including:
-- Transformer-like LLMs (e.g., Llama)
-- Mixture-of-Expert LLMs (e.g., Mixtral)
-- Embedding Models (e.g. E5-Mistral)
-- Multi-modal LLMs (e.g., LLaVA)
+vLLM seamlessly supports many Hugging Face models, including the following architectures:
 
-Find the full list of supported models [here](https://docs.vllm.ai/en/latest/models/supported_models.html).
+- Aquila & Aquila2 (`BAAI/AquilaChat2-7B`, `BAAI/AquilaChat2-34B`, `BAAI/Aquila-7B`, `BAAI/AquilaChat-7B`, etc.)
+- Baichuan & Baichuan2 (`baichuan-inc/Baichuan2-13B-Chat`, `baichuan-inc/Baichuan-7B`, etc.)
+- BLOOM (`bigscience/bloom`, `bigscience/bloomz`, etc.)
+- ChatGLM (`THUDM/chatglm2-6b`, `THUDM/chatglm3-6b`, etc.)
+- Command-R (`CohereForAI/c4ai-command-r-v01`, etc.)
+- DBRX (`databricks/dbrx-base`, `databricks/dbrx-instruct` etc.)
+- DeciLM (`Deci/DeciLM-7B`, `Deci/DeciLM-7B-instruct`, etc.)
+- Falcon (`tiiuae/falcon-7b`, `tiiuae/falcon-40b`, `tiiuae/falcon-rw-7b`, etc.)
+- Gemma (`google/gemma-2b`, `google/gemma-7b`, etc.)
+- GPT-2 (`gpt2`, `gpt2-xl`, etc.)
+- GPT BigCode (`bigcode/starcoder`, `bigcode/gpt_bigcode-santacoder`, etc.)
+- GPT-J (`EleutherAI/gpt-j-6b`, `nomic-ai/gpt4all-j`, etc.)
+- GPT-NeoX (`EleutherAI/gpt-neox-20b`, `databricks/dolly-v2-12b`, `stabilityai/stablelm-tuned-alpha-7b`, etc.)
+- InternLM (`internlm/internlm-7b`, `internlm/internlm-chat-7b`, etc.)
+- InternLM2 (`internlm/internlm2-7b`, `internlm/internlm2-chat-7b`, etc.)
+- Jais (`core42/jais-13b`, `core42/jais-13b-chat`, `core42/jais-30b-v3`, `core42/jais-30b-chat-v3`, etc.)
+- LLaMA & LLaMA-2 (`meta-llama/Llama-2-70b-hf`, `lmsys/vicuna-13b-v1.3`, `young-geng/koala`, `openlm-research/open_llama_13b`, etc.)
+- MiniCPM (`openbmb/MiniCPM-2B-sft-bf16`, `openbmb/MiniCPM-2B-dpo-bf16`, etc.)
+- Mistral (`mistralai/Mistral-7B-v0.1`, `mistralai/Mistral-7B-Instruct-v0.1`, etc.)
+- Mixtral (`mistralai/Mixtral-8x7B-v0.1`, `mistralai/Mixtral-8x7B-Instruct-v0.1`, `mistral-community/Mixtral-8x22B-v0.1`, etc.)
+- MPT (`mosaicml/mpt-7b`, `mosaicml/mpt-30b`, etc.)
+- OLMo (`allenai/OLMo-1B`, `allenai/OLMo-7B`, etc.)
+- OPT (`facebook/opt-66b`, `facebook/opt-iml-max-30b`, etc.)
+- Orion (`OrionStarAI/Orion-14B-Base`, `OrionStarAI/Orion-14B-Chat`, etc.)
+- Phi (`microsoft/phi-1_5`, `microsoft/phi-2`, etc.)
+- Qwen (`Qwen/Qwen-7B`, `Qwen/Qwen-7B-Chat`, etc.)
+- Qwen2 (`Qwen/Qwen1.5-7B`, `Qwen/Qwen1.5-7B-Chat`, etc.)
+- Qwen2MoE (`Qwen/Qwen1.5-MoE-A2.7B`, `Qwen/Qwen1.5-MoE-A2.7B-Chat`, etc.)
+- StableLM(`stabilityai/stablelm-3b-4e1t`, `stabilityai/stablelm-base-alpha-7b-v2`, etc.)
+- Starcoder2(`bigcode/starcoder2-3b`, `bigcode/starcoder2-7b`, `bigcode/starcoder2-15b`, etc.)
+- Xverse (`xverse/XVERSE-7B-Chat`, `xverse/XVERSE-13B-Chat`, `xverse/XVERSE-65B-Chat`, etc.)
+- Yi (`01-ai/Yi-6B`, `01-ai/Yi-34B`, etc.)
 
-## Getting Started
-
-Install vLLM with `pip` or [from source](https://vllm.readthedocs.io/en/latest/getting_started/installation.html#build-from-source):
+Install vLLM with pip or [from source](https://vllm.readthedocs.io/en/latest/getting_started/installation.html#build-from-source):
 
 ```bash
 pip install vllm
 ```
 
-Visit our [documentation](https://vllm.readthedocs.io/en/latest/) to learn more.
+## Getting Started
+
+Visit our [documentation](https://vllm.readthedocs.io/en/latest/) to get started.
 - [Installation](https://vllm.readthedocs.io/en/latest/getting_started/installation.html)
 - [Quickstart](https://vllm.readthedocs.io/en/latest/getting_started/quickstart.html)
 - [Supported Models](https://vllm.readthedocs.io/en/latest/models/supported_models.html)
@@ -90,36 +104,6 @@ Visit our [documentation](https://vllm.readthedocs.io/en/latest/) to learn more.
 We welcome and value any contributions and collaborations.
 Please check out [CONTRIBUTING.md](./CONTRIBUTING.md) for how to get involved.
 
-## Sponsors
-
-vLLM is a community project. Our compute resources for development and testing are supported by the following organizations. Thank you for your support!
-
-<!-- Note: Please sort them in alphabetical order. -->
-<!-- Note: Please keep these consistent with docs/source/community/sponsors.md -->
-
-- a16z
-- AMD
-- Anyscale
-- AWS
-- Crusoe Cloud
-- Databricks
-- DeepInfra
-- Dropbox
-- Google Cloud
-- Lambda Lab
-- NVIDIA
-- Replicate
-- Roblox
-- RunPod
-- Sequoia Capital
-- Skywork AI
-- Trainy
-- UC Berkeley
-- UC San Diego
-- ZhenFund
-
-We also have an official fundraising venue through [OpenCollective](https://opencollective.com/vllm). We plan to use the fund to support the development, maintenance, and adoption of vLLM.
-
 ## Citation
 
 If you use vLLM for your research, please cite our [paper](https://arxiv.org/abs/2309.06180):
@@ -131,10 +115,3 @@ If you use vLLM for your research, please cite our [paper](https://arxiv.org/abs
   year={2023}
 }
 ```
-
-## Contact Us
-
-* For technical questions and feature requests, please use Github issues or discussions.
-* For discussing with fellow users, please use Discord.
-* For security disclosures, please use Github's security advisory feature.
-* For collaborations and partnerships, please contact us at vllm-questions AT lists.berkeley.edu.

SECURITY.md

@@ -1,12 +0,0 @@
-# Security Policy
-
-## Reporting a Vulnerability
-
-If you believe you have found a security vulnerability in vLLM, we encourage you to let us know right away. 
-We will investigate all legitimate reports and do our best to quickly fix the problem.
-
-Please report security issues using https://github.com/vllm-project/vllm/security/advisories/new
-
----
-Please see PyTorch Security for more information how to securely interact with models: https://github.com/pytorch/pytorch/blob/main/SECURITY.md
-This document mostly references the recommendation from PyTorch, thank you! 

benchmarks/backend_request_func.py

@@ -4,13 +4,10 @@ import sys
 import time
 import traceback
 from dataclasses import dataclass, field
-from typing import List, Optional, Union
+from typing import List, Optional
 
 import aiohttp
-import huggingface_hub.constants
 from tqdm.asyncio import tqdm
-from transformers import (AutoTokenizer, PreTrainedTokenizer,
-                          PreTrainedTokenizerFast)
 
 AIOHTTP_TIMEOUT = aiohttp.ClientTimeout(total=6 * 60 * 60)
 
@@ -24,8 +21,6 @@ class RequestFuncInput:
     model: str
     best_of: int = 1
     use_beam_search: bool = False
-    logprobs: Optional[int] = None
-    multi_modal_content: Optional[dict] = None
 
 
 @dataclass
@@ -73,13 +68,9 @@ async def async_request_tgi(
                         chunk_bytes = chunk_bytes.strip()
                         if not chunk_bytes:
                             continue
-                        chunk_bytes = chunk_bytes.decode("utf-8")
 
-                        #NOTE: Sometimes TGI returns a ping response without
-                        # any data, we should skip it.
-                        if chunk_bytes.startswith(":"):
-                            continue
-                        chunk = remove_prefix(chunk_bytes, "data:")
+                        chunk = remove_prefix(chunk_bytes.decode("utf-8"),
+                                              "data:")
 
                         data = json.loads(chunk)
                         timestamp = time.perf_counter()
@@ -98,9 +89,6 @@ async def async_request_tgi(
                     output.latency = most_recent_timestamp - st
                     output.success = True
                     output.generated_text = data["generated_text"]
-                else:
-                    output.error = response.reason or ""
-                    output.success = False
         except Exception:
             output.success = False
             exc_info = sys.exc_info()
@@ -147,7 +135,6 @@ async def async_request_trt_llm(
                                               "data:")
 
                         data = json.loads(chunk)
-                        output.generated_text += data["text_output"]
                         timestamp = time.perf_counter()
                         # First token
                         if ttft == 0.0:
@@ -162,6 +149,7 @@ async def async_request_trt_llm(
                         most_recent_timestamp = timestamp
 
                     output.latency = most_recent_timestamp - st
+                    output.generated_text = json.loads(data)["text_output"]
                     output.success = True
 
                 else:
@@ -227,8 +215,8 @@ async def async_request_openai_completions(
 ) -> RequestFuncOutput:
     api_url = request_func_input.api_url
     assert api_url.endswith(
-        ("completions", "profile")
-    ), "OpenAI Completions API URL must end with 'completions' or 'profile'."
+        "v1/completions"
+    ), "OpenAI Completions API URL must end with 'v1/completions'."
 
     async with aiohttp.ClientSession(timeout=AIOHTTP_TIMEOUT) as session:
         assert not request_func_input.use_beam_search
@@ -238,7 +226,6 @@ async def async_request_openai_completions(
             "temperature": 0.0,
             "best_of": request_func_input.best_of,
             "max_tokens": request_func_input.output_len,
-            "logprobs": request_func_input.logprobs,
             "stream": True,
         }
         headers = {
@@ -268,9 +255,6 @@ async def async_request_openai_completions(
                         else:
                             data = json.loads(chunk)
 
-                            # NOTE: Some completion API might have a last
-                            # usage summary response without a token so we
-                            # want to check a token was generated
                             if data["choices"][0]["text"]:
                                 timestamp = time.perf_counter()
                                 # First token
@@ -279,7 +263,10 @@ async def async_request_openai_completions(
                                     output.ttft = ttft
 
                                 # Decoding phase
-                                else:
+                                # NOTE: Some completion API might have a last
+                                # usage summary response without a token so we
+                                # do not want to include as inter-token-latency
+                                elif data.get("usage", None) is None:
                                     output.itl.append(timestamp -
                                                       most_recent_timestamp)
 
@@ -289,9 +276,6 @@ async def async_request_openai_completions(
                     output.generated_text = generated_text
                     output.success = True
                     output.latency = latency
-                else:
-                    output.error = response.reason or ""
-                    output.success = False
         except Exception:
             output.success = False
             exc_info = sys.exc_info()
@@ -308,20 +292,17 @@ async def async_request_openai_chat_completions(
 ) -> RequestFuncOutput:
     api_url = request_func_input.api_url
     assert api_url.endswith(
-        "chat/completions"
-    ), "OpenAI Chat Completions API URL must end with 'chat/completions'."
+        "v1/chat/completions"
+    ), "OpenAI Chat Completions API URL must end with 'v1/chat/completions'."
 
     async with aiohttp.ClientSession(timeout=AIOHTTP_TIMEOUT) as session:
         assert not request_func_input.use_beam_search
-        content = [{"type": "text", "text": request_func_input.prompt}]
-        if request_func_input.multi_modal_content:
-            content.append(request_func_input.multi_modal_content)
         payload = {
             "model": request_func_input.model,
             "messages": [
                 {
                     "role": "user",
-                    "content": content
+                    "content": request_func_input.prompt,
                 },
             ],
             "temperature": 0.0,
@@ -397,30 +378,6 @@ def remove_prefix(text: str, prefix: str) -> str:
     return text
 
 
-def get_model(pretrained_model_name_or_path: str) -> str:
-    if os.getenv('VLLM_USE_MODELSCOPE', 'False').lower() == 'true':
-        from modelscope import snapshot_download
-
-        model_path = snapshot_download(
-            model_id=pretrained_model_name_or_path,
-            local_files_only=huggingface_hub.constants.HF_HUB_OFFLINE,
-            ignore_file_pattern=[".*.pt", ".*.safetensors", ".*.bin"])
-
-        return model_path
-    return pretrained_model_name_or_path
-
-
-def get_tokenizer(
-    pretrained_model_name_or_path: str, trust_remote_code: bool
-) -> Union[PreTrainedTokenizer, PreTrainedTokenizerFast]:
-    if pretrained_model_name_or_path is not None and not os.path.exists(
-            pretrained_model_name_or_path):
-        pretrained_model_name_or_path = get_model(
-            pretrained_model_name_or_path)
-    return AutoTokenizer.from_pretrained(pretrained_model_name_or_path,
-                                         trust_remote_code=trust_remote_code)
-
-
 ASYNC_REQUEST_FUNCS = {
     "tgi": async_request_tgi,
     "vllm": async_request_openai_completions,
@@ -429,5 +386,4 @@ ASYNC_REQUEST_FUNCS = {
     "openai": async_request_openai_completions,
     "openai-chat": async_request_openai_chat_completions,
     "tensorrt-llm": async_request_trt_llm,
-    "scalellm": async_request_openai_completions,
 }