101 lines
4.2 KiB
C++
101 lines
4.2 KiB
C++
// Copyright 2022 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 "bench/utils.h"
|
|
#include <benchmark/benchmark.h>
|
|
|
|
#include <xnnpack.h>
|
|
#include <xnnpack/aligned-allocator.h>
|
|
#include <xnnpack/common.h>
|
|
#include <xnnpack/microfnptr.h>
|
|
#include <xnnpack/window.h>
|
|
|
|
|
|
void window(
|
|
benchmark::State& state,
|
|
xnn_s16_window_ukernel_function window,
|
|
benchmark::utils::IsaCheckFunction isa_check = nullptr)
|
|
{
|
|
if (isa_check && !isa_check(state)) {
|
|
return;
|
|
}
|
|
const size_t rows = state.range(0);
|
|
const size_t batch = state.range(1);
|
|
|
|
std::vector<int16_t, AlignedAllocator<int16_t, 64>> input(
|
|
(rows * batch) + XNN_EXTRA_BYTES / sizeof(int16_t));
|
|
std::vector<int16_t, AlignedAllocator<int16_t, 64>> weights(
|
|
batch + XNN_EXTRA_BYTES / sizeof(int16_t));
|
|
std::vector<int16_t, AlignedAllocator<int16_t, 64>> output(
|
|
(rows * batch) + XNN_EXTRA_BYTES / sizeof(int16_t));
|
|
std::iota(input.begin(), input.end(), 0);
|
|
std::fill(weights.begin(), weights.end(), 0);
|
|
std::iota(output.begin(), output.end(), 0);
|
|
|
|
for (auto _ : state) {
|
|
window(rows, batch, input.data(), weights.data(), output.data(), 12 /* shift */);
|
|
}
|
|
|
|
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({"rows", "batch"});
|
|
b->Args({1, 400});
|
|
b->Args({10, 400});
|
|
}
|
|
|
|
#if XNN_ARCH_ARM || XNN_ARCH_ARM64
|
|
BENCHMARK_CAPTURE(window, s16_neon_x8, xnn_s16_window_ukernel__neon_x8, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_neon_x16, xnn_s16_window_ukernel__neon_x16, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_neon_x24, xnn_s16_window_ukernel__neon_x24, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_neon_x32, xnn_s16_window_ukernel__neon_x32, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
|
|
BENCHMARK_CAPTURE(window, s16_shift12_neon_x8, xnn_s16_window_shift12_ukernel__neon_x8, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_shift12_neon_x16, xnn_s16_window_shift12_ukernel__neon_x16, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_shift12_neon_x24, xnn_s16_window_shift12_ukernel__neon_x24, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_shift12_neon_x32, xnn_s16_window_shift12_ukernel__neon_x32, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
|
|
BENCHMARK_CAPTURE(window, s16_shift15_neon_x8, xnn_s16_window_shift15_ukernel__neon_x8, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_shift15_neon_x16, xnn_s16_window_shift15_ukernel__neon_x16, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_shift15_neon_x24, xnn_s16_window_shift15_ukernel__neon_x24, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_shift15_neon_x32, xnn_s16_window_shift15_ukernel__neon_x32, benchmark::utils::CheckNEON)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
#endif // XNN_ARCH_ARM || XNN_ARCH_ARM64
|
|
|
|
BENCHMARK_CAPTURE(window, s16_scalar_x1, xnn_s16_window_ukernel__scalar_x1)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_scalar_x2, xnn_s16_window_ukernel__scalar_x2)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_scalar_x3, xnn_s16_window_ukernel__scalar_x3)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
BENCHMARK_CAPTURE(window, s16_scalar_x4, xnn_s16_window_ukernel__scalar_x4)
|
|
->Apply(BenchmarkKernelSize)->UseRealTime();
|
|
|
|
#ifndef XNNPACK_BENCHMARK_NO_MAIN
|
|
BENCHMARK_MAIN();
|
|
#endif
|