3.9.2 doc/version (#2279)

* 3.9.2 doc/version

* whitespace
This commit is contained in:
Haicheng Wu
2025-05-04 00:00:15 -04:00
committed by GitHub
parent 40f124ef27
commit ad7b2f5e84
6 changed files with 12 additions and 6 deletions

View File

@ -1,5 +1,11 @@
# NVIDIA CUTLASS Changelog
## [3.9.2](https://github.com/NVIDIA/cutlass/releases/tag/v3.9.2) (2025-05-03)
* Fixed [Blockwise](./examples/67_hopper_fp8_warp_specialized_gemm_with_blockwise_scaling/67_hopper_fp8_warp_specialized_gemm_with_blockwise_scaling.cu) and [Groupwise](./examples/67_hopper_fp8_warp_specialized_gemm_with_blockwise_scaling/67_hopper_fp8_warp_specialized_gemm_with_groupwise_scaling.cu) GEMM hang issue when problem size K is 128.
* Optimal code generation with CUDA toolkit versions 12.9.
## [3.9.1](https://github.com/NVIDIA/cutlass/releases/tag/v3.9.1) (2025-04-30)
* Fixed Group Gemm hang issue in CUTLASS 3.x

View File

@ -1,8 +1,8 @@
![ALT](./media/images/gemm-hierarchy-with-epilogue-no-labels.png "Complete CUDA GEMM decomposition")
# CUTLASS 3.9.1
# CUTLASS 3.9.2
_CUTLASS 3.9.1 - April 2025_
_CUTLASS 3.9.2 - May 2025_
CUTLASS is a collection of CUDA C++ template abstractions for implementing
high-performance matrix-matrix multiplication (GEMM) and related computations at all levels

View File

@ -36,7 +36,7 @@
#define CUTLASS_MAJOR 3
#define CUTLASS_MINOR 9
#define CUTLASS_PATCH 1
#define CUTLASS_PATCH 2
#ifdef CUTLASS_VERSIONS_GENERATED
#include "cutlass/version_extended.h"

View File

@ -133,7 +133,7 @@ def get_option_registry():
this._option_registry = OptionRegistry(device_cc())
return this._option_registry
this.__version__ = '3.9.1'
this.__version__ = '3.9.2'
from cutlass.backend import create_memory_pool
from cutlass.emit.pytorch import pytorch

View File

@ -36,7 +36,7 @@ from setuptools import setup
def perform_setup():
setup(
name='cutlass_library',
version='3.9.1',
version='3.9.2',
description='CUTLASS library generation scripts',
packages=['cutlass_library']
)

View File

@ -36,7 +36,7 @@ from setuptools import setup
def perform_setup():
setup(
name='pycute',
version='3.9.1',
version='3.9.2',
description='Python implementation of CuTe',
packages=['pycute'],
)