Add support for mixed 4-bit/8-bit data types GEMM (#1413)
* Add support for mixed 4-bit/8-bit data types GEMM * fix ( and ) --------- Co-authored-by: Aleksandar Samardžić <asamardzic@matf.bg.ac.rs> Co-authored-by: Haicheng Wu <haichengw@nvidia.com>
This commit is contained in:
Aleksandar Samardžić
co-authored by
Aleksandar Samardžić
Haicheng Wu
Aleksandar Samardžić
Haicheng Wu
parent
f7b19de32c
commit
e1976daacc
@@ -264,6 +264,9 @@ cutlass_test_unit_add_executable(
|
||||
gemm_universal_s8t_f16n_f16t_mixed_input_tensor_op_f16_sm80.cu
|
||||
gemm_universal_u8t_f16n_f16t_mixed_input_tensor_op_f16_sm80.cu
|
||||
|
||||
gemm_universal_s4t_s8n_s32t_mixed_input_tensor_op_s32_sm80.cu
|
||||
gemm_universal_s4t_s8n_s8t_mixed_input_tensor_op_s32_sm80.cu
|
||||
|
||||
# Upcast on Operand B
|
||||
gemm_universal_f16t_s8n_f32t_mixed_input_tensor_op_f32_sm80.cu
|
||||
gemm_universal_f16t_u8n_f32t_mixed_input_tensor_op_f32_sm80.cu
|
||||
@@ -277,6 +280,9 @@ cutlass_test_unit_add_executable(
|
||||
|
||||
gemm_universal_f16t_s8n_f16t_mixed_input_tensor_op_f16_sm80.cu
|
||||
gemm_universal_f16t_u8n_f16t_mixed_input_tensor_op_f16_sm80.cu
|
||||
|
||||
gemm_universal_s8t_s4n_s32t_mixed_input_tensor_op_s32_sm80.cu
|
||||
gemm_universal_s8t_s4n_s8t_mixed_input_tensor_op_s32_sm80.cu
|
||||
)
|
||||
|
||||
cutlass_test_unit_add_executable(
|
||||
|
||||
@@ -0,0 +1,95 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2023 - 2024 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.
|
||||
*
|
||||
**************************************************************************************************/
|
||||
/*! \file
|
||||
\brief Tests for device-wide GEMM interface
|
||||
|
||||
*/
|
||||
|
||||
#include "../../common/cutlass_unit_test.h"
|
||||
#include "cutlass/cutlass.h"
|
||||
|
||||
#include "cutlass/gemm/device/gemm_universal.h"
|
||||
|
||||
#include "cutlass/util/host_tensor.h"
|
||||
#include "cutlass/util/reference/host/gemm.h"
|
||||
#include "cutlass/util/reference/host/tensor_compare.h"
|
||||
#include "cutlass/util/reference/host/tensor_copy.h"
|
||||
#include "cutlass/util/reference/host/tensor_fill.h"
|
||||
#include "cutlass/util/tensor_view_io.h"
|
||||
|
||||
#include "testbed_universal.h"
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
|
||||
TEST(SM80_Device_GemmUniversal_s4t_s8n_s32t_mixed_input_tensor_op_s32, 128x128x64_64x64x64) {
|
||||
|
||||
using ElementA = cutlass::int4b_t;
|
||||
using ElementB = int8_t;
|
||||
using ElementOutput = int32_t;
|
||||
using ElementAccumulator = int32_t;
|
||||
|
||||
using Gemm = cutlass::gemm::device::GemmUniversal<
|
||||
ElementA,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementB,
|
||||
cutlass::layout::ColumnMajor,
|
||||
ElementOutput,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementAccumulator,
|
||||
cutlass::arch::OpClassTensorOp,
|
||||
cutlass::arch::Sm80,
|
||||
cutlass::gemm::GemmShape<128, 128, 64>,
|
||||
cutlass::gemm::GemmShape<64, 64, 64>,
|
||||
cutlass::gemm::GemmShape<16, 8, 32>,
|
||||
cutlass::epilogue::thread::LinearCombination<
|
||||
ElementOutput, 128 / cutlass::sizeof_bits<ElementOutput>::value,
|
||||
ElementAccumulator, ElementAccumulator>,
|
||||
cutlass::gemm::threadblock::GemmIdentityThreadblockSwizzle<>,
|
||||
4, // Stages
|
||||
32, // AlignmentA
|
||||
16, // AlignmentB
|
||||
cutlass::arch::OpMultiplyAddMixedInputUpcast,
|
||||
cutlass::ComplexTransform::kNone,
|
||||
cutlass::ComplexTransform::kNone
|
||||
>;
|
||||
|
||||
EXPECT_TRUE(test::gemm::device::TestAllGemmUniversal<Gemm>());
|
||||
}
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#endif // #if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
@@ -0,0 +1,95 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2023 - 2024 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.
|
||||
*
|
||||
**************************************************************************************************/
|
||||
/*! \file
|
||||
\brief Tests for device-wide GEMM interface
|
||||
|
||||
*/
|
||||
|
||||
#include "../../common/cutlass_unit_test.h"
|
||||
#include "cutlass/cutlass.h"
|
||||
|
||||
#include "cutlass/gemm/device/gemm_universal.h"
|
||||
|
||||
#include "cutlass/util/host_tensor.h"
|
||||
#include "cutlass/util/reference/host/gemm.h"
|
||||
#include "cutlass/util/reference/host/tensor_compare.h"
|
||||
#include "cutlass/util/reference/host/tensor_copy.h"
|
||||
#include "cutlass/util/reference/host/tensor_fill.h"
|
||||
#include "cutlass/util/tensor_view_io.h"
|
||||
|
||||
#include "testbed_universal.h"
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
|
||||
TEST(SM80_Device_GemmUniversal_s4t_s8n_s8t_mixed_input_tensor_op_s32, 128x128x64_64x64x64) {
|
||||
|
||||
using ElementA = cutlass::int4b_t;
|
||||
using ElementB = int8_t;
|
||||
using ElementOutput = int8_t;
|
||||
using ElementAccumulator = int32_t;
|
||||
|
||||
using Gemm = cutlass::gemm::device::GemmUniversal<
|
||||
ElementA,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementB,
|
||||
cutlass::layout::ColumnMajor,
|
||||
ElementOutput,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementAccumulator,
|
||||
cutlass::arch::OpClassTensorOp,
|
||||
cutlass::arch::Sm80,
|
||||
cutlass::gemm::GemmShape<128, 128, 64>,
|
||||
cutlass::gemm::GemmShape<64, 64, 64>,
|
||||
cutlass::gemm::GemmShape<16, 8, 32>,
|
||||
cutlass::epilogue::thread::LinearCombination<
|
||||
ElementOutput, 128 / cutlass::sizeof_bits<ElementOutput>::value,
|
||||
ElementAccumulator, ElementAccumulator>,
|
||||
cutlass::gemm::threadblock::GemmIdentityThreadblockSwizzle<>,
|
||||
4, // Stages
|
||||
32, // AlignmentA
|
||||
16, // AlignmentB
|
||||
cutlass::arch::OpMultiplyAddMixedInputUpcast,
|
||||
cutlass::ComplexTransform::kNone,
|
||||
cutlass::ComplexTransform::kNone
|
||||
>;
|
||||
|
||||
EXPECT_TRUE(test::gemm::device::TestAllGemmUniversal<Gemm>());
|
||||
}
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#endif // #if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
@@ -0,0 +1,95 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2023 - 2024 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.
|
||||
*
|
||||
**************************************************************************************************/
|
||||
/*! \file
|
||||
\brief Tests for device-wide GEMM interface
|
||||
|
||||
*/
|
||||
|
||||
#include "../../common/cutlass_unit_test.h"
|
||||
#include "cutlass/cutlass.h"
|
||||
|
||||
#include "cutlass/gemm/device/gemm_universal.h"
|
||||
|
||||
#include "cutlass/util/host_tensor.h"
|
||||
#include "cutlass/util/reference/host/gemm.h"
|
||||
#include "cutlass/util/reference/host/tensor_compare.h"
|
||||
#include "cutlass/util/reference/host/tensor_copy.h"
|
||||
#include "cutlass/util/reference/host/tensor_fill.h"
|
||||
#include "cutlass/util/tensor_view_io.h"
|
||||
|
||||
#include "testbed_universal.h"
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
|
||||
TEST(SM80_Device_GemmUniversal_s8t_s4n_s32t_mixed_input_tensor_op_s32, 128x128x64_64x64x64) {
|
||||
|
||||
using ElementA = int8_t;
|
||||
using ElementB = cutlass::int4b_t;
|
||||
using ElementOutput = int32_t;
|
||||
using ElementAccumulator = int32_t;
|
||||
|
||||
using Gemm = cutlass::gemm::device::GemmUniversal<
|
||||
ElementA,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementB,
|
||||
cutlass::layout::ColumnMajor,
|
||||
ElementOutput,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementAccumulator,
|
||||
cutlass::arch::OpClassTensorOp,
|
||||
cutlass::arch::Sm80,
|
||||
cutlass::gemm::GemmShape<128, 128, 64>,
|
||||
cutlass::gemm::GemmShape<64, 64, 64>,
|
||||
cutlass::gemm::GemmShape<16, 8, 32>,
|
||||
cutlass::epilogue::thread::LinearCombination<
|
||||
ElementOutput, 128 / cutlass::sizeof_bits<ElementOutput>::value,
|
||||
ElementAccumulator, ElementAccumulator>,
|
||||
cutlass::gemm::threadblock::GemmIdentityThreadblockSwizzle<>,
|
||||
4, // Stages
|
||||
16, // AlignmentA
|
||||
32, // AlignmentB
|
||||
cutlass::arch::OpMultiplyAddMixedInputUpcast,
|
||||
cutlass::ComplexTransform::kNone,
|
||||
cutlass::ComplexTransform::kNone
|
||||
>;
|
||||
|
||||
EXPECT_TRUE(test::gemm::device::TestAllGemmUniversal<Gemm>());
|
||||
}
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#endif // #if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
@@ -0,0 +1,95 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2023 - 2024 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.
|
||||
*
|
||||
**************************************************************************************************/
|
||||
/*! \file
|
||||
\brief Tests for device-wide GEMM interface
|
||||
|
||||
*/
|
||||
|
||||
#include "../../common/cutlass_unit_test.h"
|
||||
#include "cutlass/cutlass.h"
|
||||
|
||||
#include "cutlass/gemm/device/gemm_universal.h"
|
||||
|
||||
#include "cutlass/util/host_tensor.h"
|
||||
#include "cutlass/util/reference/host/gemm.h"
|
||||
#include "cutlass/util/reference/host/tensor_compare.h"
|
||||
#include "cutlass/util/reference/host/tensor_copy.h"
|
||||
#include "cutlass/util/reference/host/tensor_fill.h"
|
||||
#include "cutlass/util/tensor_view_io.h"
|
||||
|
||||
#include "testbed_universal.h"
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
|
||||
TEST(SM80_Device_GemmUniversal_s8t_s4n_s8t_mixed_input_tensor_op_s32, 128x128x64_64x64x64) {
|
||||
|
||||
using ElementA = int8_t;
|
||||
using ElementB = cutlass::int4b_t;
|
||||
using ElementOutput = int8_t;
|
||||
using ElementAccumulator = int32_t;
|
||||
|
||||
using Gemm = cutlass::gemm::device::GemmUniversal<
|
||||
ElementA,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementB,
|
||||
cutlass::layout::ColumnMajor,
|
||||
ElementOutput,
|
||||
cutlass::layout::RowMajor,
|
||||
ElementAccumulator,
|
||||
cutlass::arch::OpClassTensorOp,
|
||||
cutlass::arch::Sm80,
|
||||
cutlass::gemm::GemmShape<128, 128, 64>,
|
||||
cutlass::gemm::GemmShape<64, 64, 64>,
|
||||
cutlass::gemm::GemmShape<16, 8, 32>,
|
||||
cutlass::epilogue::thread::LinearCombination<
|
||||
ElementOutput, 128 / cutlass::sizeof_bits<ElementOutput>::value,
|
||||
ElementAccumulator, ElementAccumulator>,
|
||||
cutlass::gemm::threadblock::GemmIdentityThreadblockSwizzle<>,
|
||||
4, // Stages
|
||||
16, // AlignmentA
|
||||
32, // AlignmentB
|
||||
cutlass::arch::OpMultiplyAddMixedInputUpcast,
|
||||
cutlass::ComplexTransform::kNone,
|
||||
cutlass::ComplexTransform::kNone
|
||||
>;
|
||||
|
||||
EXPECT_TRUE(test::gemm::device::TestAllGemmUniversal<Gemm>());
|
||||
}
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
#endif // #if defined(CUTLASS_ARCH_MMA_SM80_SUPPORTED)
|
||||
|
||||
////////////////////////////////////////////////////////////////////////////////
|
||||
Reference in New Issue
Block a user