brintos

brintos / llvm-project-archived public Read only

0
0
Text · 2.9 KiB · 2398c4b Raw
60 lines · cpp
1//===-- GPU benchmark for exp ---------------------------------------------===//2//3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.4// See https://llvm.org/LICENSE.txt for license information.5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception6//7//===----------------------------------------------------------------------===//8 9#include "benchmarks/gpu/LibcGpuBenchmark.h"10#include "benchmarks/gpu/Random.h"11 12#include "hdr/stdint_proxy.h"13#include "src/math/exp.h"14 15#if defined(NVPTX_MATH_FOUND) || defined(AMDGPU_MATH_FOUND)16#include "platform.h"17#endif18 19#define RANDOM_INPUT(T, Func, Dist, Min, Max, N)                               \20  [](uint32_t call_index) {                                                    \21    using namespace LIBC_NAMESPACE::benchmarks;                                \22                                                                               \23    const Dist<T> dist(Min, Max);                                              \24    return MathPerf<T>::template run_throughput<N>(Func, dist, call_index);    \25  }26 27#define BENCH(T, Name, Func, Dist, Min, Max)                                   \28  SINGLE_WAVE_BENCHMARK(LlvmLibcExpGpuBenchmark, Name##_1,                     \29                        RANDOM_INPUT(T, Func, Dist, Min, Max, 1));             \30  SINGLE_WAVE_BENCHMARK(LlvmLibcExpGpuBenchmark, Name##_128,                   \31                        RANDOM_INPUT(T, Func, Dist, Min, Max, 128));           \32  SINGLE_WAVE_BENCHMARK(LlvmLibcExpGpuBenchmark, Name##_1024,                  \33                        RANDOM_INPUT(T, Func, Dist, Min, Max, 1024));          \34  SINGLE_WAVE_BENCHMARK(LlvmLibcExpGpuBenchmark, Name##_4096,                  \35                        RANDOM_INPUT(T, Func, Dist, Min, Max, 4096))36 37using LIBC_NAMESPACE::exp;38 39BENCH(double, ExpSubnormal, exp, UniformExponent, -1022, -1022);40BENCH(double, ExpCoreRange, exp, UniformLinear, -10.0, 10.0);41BENCH(double, ExpFinite, exp, UniformLinear, -745.0, 709.0);42BENCH(double, ExpUnderflow, exp, UniformLinear, -746.0, -745.0);43BENCH(double, ExpOverflow, exp, UniformLinear, 709.0, 710.0);44 45#ifdef NVPTX_MATH_FOUND46BENCH(double, NvExpSubnormal, __nv_exp, UniformExponent, -1022, -1022);47BENCH(double, NvExpCoreRange, __nv_exp, UniformLinear, -10.0, 10.0);48BENCH(double, NvExpFinite, __nv_exp, UniformLinear, -745.0, 709.0);49BENCH(double, NvExpUnderflow, __nv_exp, UniformLinear, -746.0, -745.0);50BENCH(double, NvExpOverflow, __nv_exp, UniformLinear, 709.0, 710.0);51#endif52 53#ifdef AMDGPU_MATH_FOUND54BENCH(double, AmdExpSubnormal, __ocml_exp_f64, UniformExponent, -1022, -1022);55BENCH(double, AmdExpCoreRange, __ocml_exp_f64, UniformLinear, -10.0, 10.0);56BENCH(double, AmdExpFinite, __ocml_exp_f64, UniformLinear, -745.0, 709.0);57BENCH(double, AmdExpUnderflow, __ocml_exp_f64, UniformLinear, -746.0, -745.0);58BENCH(double, AmdExpOverflow, __ocml_exp_f64, UniformLinear, 709.0, 710.0);59#endif60