Files
cutlass/test/python/cutlass/gemm/gemm_mixed_sm80.py
Junkai-Wu a49a78ffef v4.2 release. (#2587)
* Fix default cluster callback values to 1 to avoid profiler failure when these values are not set in command line.

* v4.2 release.
2025-08-22 18:11:24 -04:00

76 lines
3.3 KiB
Python

#################################################################################################
#
# Copyright (c) 2023 - 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: BSD-3-Clause
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
#################################################################################################
"""
Low-level functionality tests for GEMM with mixed operands on SM80
"""
from functools import partial
import logging
import unittest
import cutlass_cppgen
from cutlass_cppgen.backend.utils.device import device_cc
from utils import LayoutCombination, add_test_gemm
cutlass_cppgen.set_log_level(logging.WARNING)
cc = 80
dtype =cutlass_cppgen.DataType.f16
@unittest.skipIf(device_cc() < cc, 'Device compute capability is insufficient for SM80 tests.')
@unittest.skipIf(cutlass_cppgen.utils.datatypes.torch_type(dtype) is None, f'Version of torch installed does not contain a datatype match for {dtype}')
class GemmMixedSm80(unittest.TestCase):
"""
Wrapper class to which tests will be added dynamically in __main__
"""
pass
add_test_mixed = partial(add_test_gemm, cls=GemmMixedSm80, element=dtype, cc=cc, cluster_shape=[1, 1, 1],
opclass=cutlass_cppgen.OpcodeClass.TensorOp, threadblock_shape=[128, 128, 64],
warp_count=[2, 2, 1], stages=3, element_accumulator=cutlass_cppgen.DataType.f32)
# Test with upcast on A
add_test_mixed(element_A=cutlass_cppgen.DataType.s8, alignments=[16, 8, 8], layouts=LayoutCombination.TNT)
add_test_mixed(element_A=cutlass_cppgen.DataType.s8, alignments=[16, 8, 8], layouts=LayoutCombination.TNN)
# Test with upcast on B
add_test_mixed(element_B=cutlass_cppgen.DataType.s8, alignments=[8, 16, 8], layouts=LayoutCombination.TNT)
add_test_mixed(element_B=cutlass_cppgen.DataType.s8, alignments=[8, 16, 8], layouts=LayoutCombination.TNN)
if __name__ == '__main__':
unittest.main()