Add Ampere half-float benchmark (#66)

Add a GEMM benchmark for ampere where the input is half and the output is float. --------- Co-authored-by: Mehdi Goli <[email protected]>
codeplaysoftware · May 30, 2024 · 0e71bd2 · 0e71bd2
1 parent 2b0e930
commit 0e71bd2
Show file tree

Hide file tree

Showing 7 changed files with 331 additions and 2 deletions.
diff --git a/benchmarks/CMakeLists.txt b/benchmarks/CMakeLists.txt
@@ -67,3 +67,6 @@ endfunction()
 if(SYCL_INTEL_TARGET)
   add_subdirectory(pvc)
 endif()
+if (SYCL_NVIDIA_TARGET)
+  add_subdirectory(ampere)
+endif()
diff --git a/benchmarks/ampere/CMakeLists.txt b/benchmarks/ampere/CMakeLists.txt
@@ -0,0 +1,33 @@
+# Copyright (c) 2024 - 2024 Codeplay Software Ltd. 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.
+
+
+cutlass_benchmark_add_executable(
+  bench_ampere_gemm_fp16_fp16_fp32_tensor_op_fp32
+  bench_ampere_gemm_fp16_fp16_fp32_tensor_op_fp32.cpp
+)
diff --git a/benchmarks/ampere/bench_ampere_gemm_fp16_fp16_fp32_tensor_op_fp32.cpp b/benchmarks/ampere/bench_ampere_gemm_fp16_fp16_fp32_tensor_op_fp32.cpp
@@ -0,0 +1,153 @@
+/***************************************************************************************************
+ * Copyright (c) 2024 - 2024 Codeplay Software Ltd. 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.
+ *
+ **************************************************************************************************/
+
+#include "../common/benchmark_runner.hpp"
+#include "gemm_configuration.hpp"
+
+int main(int argc, const char** argv)
+{
+  //
+  // Parse options
+  //
+
+  Options options;
+
+  options.parse(argc, argv);
+
+  if (options.help) {
+    options.print_usage(std::cout) << std::endl;
+    return 0;
+  }
+
+  if (options.error) {
+    std::cerr << "Aborting execution." << std::endl;
+    return -1;
+  }
+
+  //
+  // Run Benchmark 
+  //
+
+  // The KernelHardwareInfo struct holds the number of EUs on the GPU with a given device ID. This
+  // information is used by the underlying kernel.
+  cutlass::KernelHardwareInfo hw_info;
+
+  // Change device_id to another value if you are running on a machine with multiple GPUs and wish
+  // to use a GPU other than that with device ID 0.
+  hw_info.sm_count = cutlass::KernelHardwareInfo::query_device_multiprocessor_count(hw_info.device_id);
+
+// The code section below describes datatype for input, output matrices and computation between
+// elements in input matrices.
+  using ElementAccumulator = float;                   // <- data type of accumulator
+  using ElementComputeEpilogue = float;  // <- data type of epilogue operations
+  using ElementInputA = half_t;                        // <- data type of elements in input matrix A
+  using ElementInputB = half_t;                        // <- data type of elements in input matrix B
+  using ElementOutput = float;                        // <- data type of elements in output matrix D
+
+  using LayoutA = cutlass::layout::ColumnMajor;
+  using LayoutB = cutlass::layout::ColumnMajor;
+  using LayoutC = cutlass::layout::ColumnMajor;
+  using LayoutD = cutlass::layout::ColumnMajor;
+
+  using TileShape = Shape<_128, _128, _32>;
+
+  using TiledMma = TiledMMA<
+          MMA_Atom<SM80_16x8x16_F32F16F16F32_TN>,
+          Layout<Shape<_2,_2,_1>>, // 2x2x1 thread group
+          Tile<_32,_32,_16>>;                           // 32x32x8 MMA for LDSM, 1x2x1 value group
+
+  static constexpr int kAlignmentA = 8;
+  using DefaultOperandA = DefaultGemm_TensorOpSm80_OperandA<
+          ElementInputA, LayoutA, kAlignmentA, 32>;
+  using SmemLayoutAtomA = typename DefaultOperandA::SmemLayoutAtom; // M, K
+  using SmemCopyAtomA = typename DefaultOperandA::SmemCopyAtom;
+  using GmemTiledCopyA = typename DefaultOperandA::GmemTiledCopy;
+
+  static constexpr int kAlignmentB = 8;
+  using DefaultOperandB = DefaultGemm_TensorOpSm80_OperandB<
+          ElementInputB, LayoutB, kAlignmentB, 32>;
+  using SmemLayoutAtomB = typename DefaultOperandB::SmemLayoutAtom; // N, K
+  using SmemCopyAtomB = typename DefaultOperandB::SmemCopyAtom;
+  using GmemTiledCopyB = typename DefaultOperandB::GmemTiledCopy;
+
+  using Stages = Int<3>;
+
+  // This code section describes the epilogue part of the kernel
+  using EpilogueOp = cutlass::epilogue::thread::LinearCombination<
+          ElementOutput,                                     // <- data type of output matrix
+          128 / cutlass::sizeof_bits<ElementOutput>::value,  // <- the number of elements per vectorized
+          // memory access. For a byte, it's 16
+          // elements. This becomes the vector width of
+          // math instructions in the epilogue too
+          ElementAccumulator,                                // <- data type of accumulator
+          ElementComputeEpilogue>;  // <- data type for alpha/beta in linear combination function
+
+  using DispatchPolicy = cutlass::gemm::MainloopSm80CpAsync<Stages{}>;
+
+  // Define strides (mixed)
+  using StrideA = cutlass::detail::TagToStrideA_t<LayoutA>;
+  using StrideB = cutlass::detail::TagToStrideB_t<LayoutB>;
+  using StrideC = cutlass::detail::TagToStrideC_t<LayoutC>;
+  using StrideD = cutlass::detail::TagToStrideC_t<LayoutD>;
+
+  using CollectiveEpilogue = cutlass::epilogue::collective::DefaultEpilogue<
+          StrideC,
+          StrideD,
+          EpilogueOp,
+          cutlass::gemm::EpilogueDefault>;
+
+  // Mainloop
+  using CollectiveMainloop = cutlass::gemm::collective::CollectiveMma<
+          DispatchPolicy,
+          TileShape,
+          ElementInputA,
+          StrideA,
+          ElementInputB,
+          StrideB,
+          TiledMma,
+          GmemTiledCopyA, SmemLayoutAtomA, SmemCopyAtomA, cute::identity,  // A
+          GmemTiledCopyB, SmemLayoutAtomB, SmemCopyAtomB, cute::identity   // B
+  >;
+
+  using GemmKernel = cutlass::gemm::kernel::GemmUniversal<
+          Shape<int, int, int, int>,
+          CollectiveMainloop,
+          CollectiveEpilogue
+  >;
+
+  using Gemm = cutlass::gemm::device::GemmUniversalAdapter<GemmKernel>;
+
+  BenchmarkRunner<Gemm> runner;
+
+  runner.run(options, hw_info);
+
+  return 0;
+}
diff --git a/benchmarks/ampere/gemm_configuration.hpp b/benchmarks/ampere/gemm_configuration.hpp
@@ -0,0 +1,122 @@
+/***************************************************************************************************
+ * Copyright (c) 2024 - 2024 Codeplay Software Ltd. 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.
+ *
+ **************************************************************************************************/
+
+#include "cutlass/half.h"
+#include "cutlass/layout/layout.h"
+
+#include "cute/swizzle.hpp"
+#include "cute/layout.hpp"
+#include "cute/arch/copy_sm75.hpp"
+#include "cute/arch/copy_sm80.hpp"
+#include "cute/atom/copy_atom.hpp"
+
+using namespace cute;
+
+template <typename Element, typename Layout, int Alignment, int SizeK>
+struct DefaultGemm_TensorOpSm80_OperandA;
+
+template <typename Element, typename Layout, int Alignment, int SizeK>
+struct DefaultGemm_TensorOpSm80_OperandB;
+
+/////////////////////////////////////////////////////////////////////////
+
+// half
+
+/// Operand A - Row-major (K-Major)
+template <>
+struct DefaultGemm_TensorOpSm80_OperandA<cutlass::half_t, cutlass::layout::RowMajor, 8, 64>
+{
+    // Smem
+    using SmemLayoutAtom = decltype(
+    composition(Swizzle<3,3,3>{},
+                Layout<Shape < _8,_64>,
+                Stride<_64, _1>>{}));
+    using SmemCopyAtom = Copy_Atom<SM75_U32x4_LDSM_N, half_t>;
+
+    // Gmem
+    using GmemTiledCopy = decltype(
+    make_tiled_copy(Copy_Atom<SM80_CP_ASYNC_CACHEGLOBAL<cute::uint128_t>, half_t>{},
+                    Layout<Shape <_16,_8>,
+                    Stride< _8,_1>>{},
+                    Layout<Shape < _1,_8>>{}));
+};
+
+/// Operand A - Column-major (M-major)
+template <int SizeK>
+struct DefaultGemm_TensorOpSm80_OperandA<half_t, cutlass::layout::ColumnMajor, 8, SizeK>
+{
+    // Smem
+    using SmemLayoutAtom = decltype(
+    composition(Swizzle<3,3,3>{},
+                Layout<Shape <_64, _8>,
+                Stride< _1,_64>>{}));
+    using SmemCopyAtom = Copy_Atom<SM75_U16x8_LDSM_T, half_t>;
+
+    // Gmem
+    using GmemTiledCopy = decltype(
+    make_tiled_copy(Copy_Atom<SM80_CP_ASYNC_CACHEGLOBAL<cute::uint128_t>, half_t>{},
+                    Layout<Shape <_16, _8>,
+                    Stride< _1,_16>>{},
+                    Layout<Shape < _8, _1>>{}));
+};
+
+/// Operand A - Row-major (K-Major)
+template <>
+struct DefaultGemm_TensorOpSm80_OperandA<half_t, cutlass::layout::RowMajor, 8, 32>
+{
+    // Smem
+    using SmemLayoutAtom = decltype(
+    composition(Swizzle<2,3,3>{},
+                Layout<Shape < _8,_32>,
+                Stride<_32, _1>>{}));
+    using SmemCopyAtom = Copy_Atom<SM75_U32x4_LDSM_N, half_t>;
+
+    // Gmem
+    using GmemTiledCopy = decltype(
+    make_tiled_copy(Copy_Atom<SM80_CP_ASYNC_CACHEGLOBAL<cute::uint128_t>, half_t>{},
+                    Layout<Shape <_32,_4>,
+                    Stride< _4,_1>>{},
+                    Layout<Shape < _1,_8>>{}));
+};
+
+// Because the F32F16 TiledMMA is A-B symmetric, we can reuse the DefaultOperands
+
+// Operand B - Column-Major (K-major)
+template <int Alignment, int SizeK>
+struct DefaultGemm_TensorOpSm80_OperandB<half_t, cutlass::layout::ColumnMajor, Alignment, SizeK>
+        : DefaultGemm_TensorOpSm80_OperandA<half_t, cutlass::layout::RowMajor, Alignment, SizeK>
+{};
+
+// Operand B - Row-Major (N-major)
+template <int Alignment, int SizeK>
+struct DefaultGemm_TensorOpSm80_OperandB<half_t, cutlass::layout::RowMajor, Alignment, SizeK>
+        : DefaultGemm_TensorOpSm80_OperandA<half_t, cutlass::layout::ColumnMajor, Alignment, SizeK>
+{};
diff --git a/benchmarks/common/benchmark_runner.hpp b/benchmarks/common/benchmark_runner.hpp
@@ -53,7 +53,7 @@ template <typename T>
 static void fill_matrix(std::vector<T> &M)
 {
   std::generate(std::begin(M), std::end(M), [&]
-  { return static_cast<T>( (rand() / double(RAND_MAX)) ); });
+  { return static_cast<T>( 2 * (rand() / double(RAND_MAX)) - 1); });
 }
 
 using namespace cute;

diff --git a/include/cute/arch/copy_sm75.hpp b/include/cute/arch/copy_sm75.hpp
@@ -48,6 +48,10 @@
   #define CUTE_ARCH_NVCC_SUPPORTS_LDSM_SM75 ((__CUDACC_VER_MAJOR__  == 10 && __CUDACC_VER_MINOR__ >= 2) || __CUDACC_VER_MAJOR__ >= 11)
 #endif
 
+#if defined(SYCL_NVIDIA_TARGET)
+  #define CUTE_ARCH_NVCC_SUPPORTS_LDSM_SM75 1
+#endif
+
 #if ! defined(CUTE_ARCH_LDSM_SM75_SUPPORTED)
   #define CUTE_ARCH_LDSM_SM75_SUPPORTED (CUTE_ARCH_NVCC_SUPPORTS_LDSM_SM75 || CUTE_ARCH_CLANG_SUPPORTS_LDSM_SM75)
 #endif

diff --git a/include/cutlass/vector_types.h b/include/cutlass/vector_types.h
@@ -31,13 +31,17 @@
 #pragma once
 
 #if defined(CUTLASS_ENABLE_SYCL)
-#include <sycl/sycl.hpp>
+#include "cutlass/detail/helper_macros.hpp"
 
 // Add these definitions in the cutlass namespace, so they do not clash with the ones in cuda
 namespace cutlass {
     // We use this struct instead of sycl::int4 because the sycl version requires x() to access x,
     // while the struct does not need the (). This prevents us from having to modify the Cutlass
     // implementation in all the places where these vector types are used.
+    using int2 = struct alignas(8) {
+        int x, y;
+    };
+
     using int4 = struct alignas(16) {
         int x, y, z, w;
     };
@@ -85,6 +89,16 @@ namespace cutlass {
     using double4 = struct alignas(16) {
         long long int x, y, z, w;
     };
+
+    CUTLASS_HOST_DEVICE
+    int2 make_int2(int x, int y) {
+      return int2{x,y};
+    }
+
+    CUTLASS_HOST_DEVICE
+    int4 make_int4(int x, int y, int z, int w) {
+      return int4 {x,y,z,w};
+    }
 }
 #else
 #include <vector_types.h>