sglang_v0.5.2/pytorch_2.8.0/third_party/XNNPACK/bench/xN-transposec.cc

78 lines
2.5 KiB
C++

// Copyright 2021 Google LLC
//
// This source code is licensed under the BSD-style license found in the
// LICENSE file in the root directory of this source tree.
#include <algorithm>
#include <cmath>
#include <functional>
#include <numeric>
#include <vector>
#include "utils.h"
#include "xnnpack.h"
#include "xnnpack/common.h"
#include "xnnpack/hardware-config.h"
#include "xnnpack/microfnptr.h"
#include "xnnpack/microparams-init.h"
#include "xnnpack/transpose.h"
#include "xnnpack/buffer.h"
#include <benchmark/benchmark.h>
void transpose(benchmark::State& state, uint64_t arch_flags,
xnn_transposec_ukernel_fn ukernel,
size_t element_size_bits) {
if (!benchmark::utils::CheckArchFlags(state, arch_flags)) {
return;
}
const size_t element_size = element_size_bits / 8;
const size_t height = state.range(0);
const size_t width = state.range(1);
const size_t tile_hbytes = height * element_size * sizeof(uint8_t);
const size_t tile_wbytes = width * element_size * sizeof(uint8_t);
xnnpack::Buffer<uint8_t, XNN_ALLOCATION_ALIGNMENT> x(
height * width * element_size + XNN_EXTRA_BYTES / sizeof(uint8_t));
xnnpack::Buffer<uint8_t, XNN_ALLOCATION_ALIGNMENT> y(
height * width * element_size + XNN_EXTRA_BYTES / sizeof(uint8_t));
std::iota(x.begin(), x.end(), 0);
for (auto _ : state) {
ukernel(x.data(), y.data(), tile_wbytes, tile_hbytes, width, height);
}
const uint64_t cpu_frequency = benchmark::utils::GetCurrentCpuFrequency();
if (cpu_frequency != 0) {
state.counters["cpufreq"] = cpu_frequency;
}
}
static void BenchmarkKernelSize(benchmark::internal::Benchmark* b)
{
b->ArgNames({"height", "width"});
b->Args({32, 32});
b->Args({64, 64});
b->Args({117, 117});
b->Args({128, 128});
b->Args({256, 256});
b->Args({512, 512});
b->Args({1024, 1024});
}
#define XNN_TRANSPOSE_UKERNEL(arch_flags, ukernel, element_size, ...) \
BENCHMARK_CAPTURE(transpose, ukernel, arch_flags, \
reinterpret_cast<xnn_transposec_ukernel_fn>(ukernel), \
element_size) \
->Apply(BenchmarkKernelSize) \
->UseRealTime();
#include "x8-transposec/x8-transposec.h"
#include "x16-transposec/x16-transposec.h"
#include "x24-transposec/x24-transposec.h"
#include "x32-transposec/x32-transposec.h"
#include "x64-transposec/x64-transposec.h"
#undef XNN_TRANSPOSE_UKERNEL
#ifndef XNNPACK_BENCHMARK_NO_MAIN
BENCHMARK_MAIN();
#endif