Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add vectorized f32-vlog microkernels. #6614

Merged
merged 1 commit into from
Jun 28, 2024
Merged

Add vectorized f32-vlog microkernels. #6614

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions BUILD.bazel
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -250,6 +250,7 @@ MICROKERNEL_DEPS = [

SIMD_HEADERS = ["src/xnnpack/simd/f32-" + arch + ".h" for arch in xnnpack_simd_archs()] + [
"src/xnnpack/simd/f32-avx-base.h",
"src/xnnpack/simd/f32-generic-functions.h",
]

exports_files(SIMD_HEADERS)
Expand Down
1 change: 1 addition & 0 deletions bench/BUILD.bazel
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -548,6 +548,7 @@ xnnpack_benchmark(
"f32_vabs",
"f32_velu",
"f32_vhswish",
"f32_vlog",
"f32_vlrelu",
"f32_vneg",
"f32_vrelu",
Expand Down
219 changes: 219 additions & 0 deletions bench/f32-vlog.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,219 @@
// Copyright 2024 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.
//
// Auto-generated file. Do not edit!
// Specification: test/f32-vlog.yaml
// Generator: tools/generate-vunary-benchmark.py

#include <stddef.h>
#include <stdint.h>

#include "xnnpack.h"
#include "xnnpack/aligned-allocator.h"
#include "xnnpack/common.h"
#include "xnnpack/microfnptr.h"
#include "xnnpack/microparams-init.h"
#include "xnnpack/microparams.h"
#include "xnnpack/vunary.h"

#include "bench/f32-vunary-benchmark.h"
#include "bench/utils.h"
#include <benchmark/benchmark.h>

void f32_vlog(benchmark::State& state, xnn_f32_vlog_ukernel_fn ukernel,
xnn_init_f32_default_params_fn init_params = nullptr,
benchmark::utils::IsaCheckFunction isa_check = nullptr) {
f32_vunary_benchmark<xnn_f32_default_params>(
state, ukernel,
init_params,
isa_check,
/*range_min=*/0.0,
/*range_max=*/10.0);
}

BENCHMARK_CAPTURE(f32_vlog, scalar_log_u1,
xnn_f32_vlog_ukernel__scalar_log_u1,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, scalar_log_u2,
xnn_f32_vlog_ukernel__scalar_log_u2,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, scalar_log_u4,
xnn_f32_vlog_ukernel__scalar_log_u4,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, scalar_rational_3_3_div_u1,
xnn_f32_vlog_ukernel__scalar_rational_3_3_div_u1,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, scalar_rational_3_3_div_u2,
xnn_f32_vlog_ukernel__scalar_rational_3_3_div_u2,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, scalar_rational_3_3_div_u4,
xnn_f32_vlog_ukernel__scalar_rational_3_3_div_u4,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, scalar_rational_3_3_div_u8,
xnn_f32_vlog_ukernel__scalar_rational_3_3_div_u8,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
#if XNN_ARCH_X86 || XNN_ARCH_X86_64
BENCHMARK_CAPTURE(f32_vlog, sse2_rational_3_3_div_u4,
xnn_f32_vlog_ukernel__sse2_rational_3_3_div_u4,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, sse2_rational_3_3_div_u8,
xnn_f32_vlog_ukernel__sse2_rational_3_3_div_u8,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, sse2_rational_3_3_div_u12,
xnn_f32_vlog_ukernel__sse2_rational_3_3_div_u12,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, sse2_rational_3_3_div_u16,
xnn_f32_vlog_ukernel__sse2_rational_3_3_div_u16,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx2_rational_3_3_div_u8,
xnn_f32_vlog_ukernel__avx2_rational_3_3_div_u8,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX2)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx2_rational_3_3_div_u16,
xnn_f32_vlog_ukernel__avx2_rational_3_3_div_u16,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX2)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx2_rational_3_3_div_u24,
xnn_f32_vlog_ukernel__avx2_rational_3_3_div_u24,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX2)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx2_rational_3_3_div_u32,
xnn_f32_vlog_ukernel__avx2_rational_3_3_div_u32,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX2)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, fma3_rational_3_3_div_u8,
xnn_f32_vlog_ukernel__fma3_rational_3_3_div_u8,
/*init_params=*/nullptr,
benchmark::utils::CheckFMA3)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, fma3_rational_3_3_div_u16,
xnn_f32_vlog_ukernel__fma3_rational_3_3_div_u16,
/*init_params=*/nullptr,
benchmark::utils::CheckFMA3)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, fma3_rational_3_3_div_u24,
xnn_f32_vlog_ukernel__fma3_rational_3_3_div_u24,
/*init_params=*/nullptr,
benchmark::utils::CheckFMA3)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, fma3_rational_3_3_div_u32,
xnn_f32_vlog_ukernel__fma3_rational_3_3_div_u32,
/*init_params=*/nullptr,
benchmark::utils::CheckFMA3)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx512f_rational_3_3_div_u16,
xnn_f32_vlog_ukernel__avx512f_rational_3_3_div_u16,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX512F)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx512f_rational_3_3_div_u32,
xnn_f32_vlog_ukernel__avx512f_rational_3_3_div_u32,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX512F)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx512f_rational_3_3_div_u48,
xnn_f32_vlog_ukernel__avx512f_rational_3_3_div_u48,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX512F)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, avx512f_rational_3_3_div_u64,
xnn_f32_vlog_ukernel__avx512f_rational_3_3_div_u64,
/*init_params=*/nullptr,
benchmark::utils::CheckAVX512F)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
#endif // XNN_ARCH_X86 || XNN_ARCH_X86_64

#if XNN_ARCH_ARM || XNN_ARCH_ARM64
BENCHMARK_CAPTURE(f32_vlog, neon_rational_3_3_div_u4,
xnn_f32_vlog_ukernel__neon_rational_3_3_div_u4,
/*init_params=*/nullptr,
benchmark::utils::CheckNEON)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, neon_rational_3_3_div_u8,
xnn_f32_vlog_ukernel__neon_rational_3_3_div_u8,
/*init_params=*/nullptr,
benchmark::utils::CheckNEON)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, neon_rational_3_3_div_u12,
xnn_f32_vlog_ukernel__neon_rational_3_3_div_u12,
/*init_params=*/nullptr,
benchmark::utils::CheckNEON)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, neon_rational_3_3_div_u16,
xnn_f32_vlog_ukernel__neon_rational_3_3_div_u16,
/*init_params=*/nullptr,
benchmark::utils::CheckNEON)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
#endif // XNN_ARCH_ARM || XNN_ARCH_ARM64

#if XNN_ARCH_WASMSIMD || XNN_ARCH_WASMRELAXEDSIMD
BENCHMARK_CAPTURE(f32_vlog, wasmsimd_rational_3_3_div_u4,
xnn_f32_vlog_ukernel__wasmsimd_rational_3_3_div_u4,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, wasmsimd_rational_3_3_div_u8,
xnn_f32_vlog_ukernel__wasmsimd_rational_3_3_div_u8,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, wasmsimd_rational_3_3_div_u12,
xnn_f32_vlog_ukernel__wasmsimd_rational_3_3_div_u12,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
BENCHMARK_CAPTURE(f32_vlog, wasmsimd_rational_3_3_div_u16,
xnn_f32_vlog_ukernel__wasmsimd_rational_3_3_div_u16,
/*init_params=*/nullptr)
->Apply(benchmark::utils::UnaryElementwiseParameters<float, float>)
->UseRealTime();
#endif // XNN_ARCH_WASMSIMD || XNN_ARCH_WASMRELAXEDSIMD


#ifndef XNNPACK_BENCHMARK_NO_MAIN
BENCHMARK_MAIN();
#endif
12 changes: 6 additions & 6 deletions bench/qs8-rsum.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,12 +11,12 @@
#include "bench/utils.h"
#include <benchmark/benchmark.h>

#include <xnnpack.h>
#include <xnnpack/aligned-allocator.h>
#include <xnnpack/common.h>
#include <xnnpack/reduce.h>
#include <xnnpack/microfnptr.h>
#include <xnnpack/microparams-init.h>
#include "xnnpack.h"
#include "xnnpack/aligned-allocator.h"
#include "xnnpack/common.h"
#include "xnnpack/reduce.h"
#include "xnnpack/microfnptr.h"
#include "xnnpack/microparams-init.h"


BENCHMARK_CAPTURE(qs8_rsum, scalar_u1,
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/avx2_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -230,6 +230,7 @@ SET(ALL_AVX2_MICROKERNEL_SRCS
src/f32-velu/gen/f32-velu-avx2-rr1-p6-u64.c
src/f32-velu/gen/f32-velu-avx2-rr1-p6-u72.c
src/f32-velu/gen/f32-velu-avx2-rr1-p6-u80.c
src/f32-vlog/gen/f32-vlog-avx2-rational-3-3-div.c
src/f32-vscaleexpminusmax/gen/f32-vscaleexpminusmax-avx2-p5-u8.c
src/f32-vscaleexpminusmax/gen/f32-vscaleexpminusmax-avx2-p5-u16.c
src/f32-vscaleexpminusmax/gen/f32-vscaleexpminusmax-avx2-p5-u24.c
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/avx512f_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -171,6 +171,7 @@ SET(ALL_AVX512F_MICROKERNEL_SRCS
src/f32-velu/gen/f32-velu-avx512f-rr1-p6-u128.c
src/f32-vhswish/gen/f32-vhswish-avx512f-u16.c
src/f32-vhswish/gen/f32-vhswish-avx512f-u32.c
src/f32-vlog/gen/f32-vlog-avx512f-rational-3-3-div.c
src/f32-vlrelu/gen/f32-vlrelu-avx512f-u16.c
src/f32-vlrelu/gen/f32-vlrelu-avx512f-u32.c
src/f32-vrelu/gen/f32-vrelu-avx512f-u16.c
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/fma3_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -182,6 +182,7 @@ SET(ALL_FMA3_MICROKERNEL_SRCS
src/f32-qc8w-gemm/gen/f32-qc8w-gemm-8x16-minmax-fma3-broadcast.c
src/f32-vhswish/gen/f32-vhswish-fma3-u8.c
src/f32-vhswish/gen/f32-vhswish-fma3-u16.c
src/f32-vlog/gen/f32-vlog-fma3-rational-3-3-div.c
src/f32-vrsqrt/gen/f32-vrsqrt-fma3-rsqrt-u8.c
src/f32-vrsqrt/gen/f32-vrsqrt-fma3-rsqrt-u16.c
src/f32-vrsqrt/gen/f32-vrsqrt-fma3-rsqrt-u32.c
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/neon_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -344,6 +344,7 @@ SET(ALL_NEON_MICROKERNEL_SRCS
src/f32-vhswish/gen/f32-vhswish-neon-u4.c
src/f32-vhswish/gen/f32-vhswish-neon-u8.c
src/f32-vhswish/gen/f32-vhswish-neon-u16.c
src/f32-vlog/gen/f32-vlog-neon-rational-3-3-div.c
src/f32-vlrelu/gen/f32-vlrelu-neon-u4.c
src/f32-vlrelu/gen/f32-vlrelu-neon-u8.c
src/f32-vmulcaddc/gen/f32-vmulcaddc-c4-minmax-neon-2x.c
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/scalar_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -458,6 +458,7 @@ SET(ALL_SCALAR_MICROKERNEL_SRCS
src/f32-vhswish/gen/f32-vhswish-scalar-u2.c
src/f32-vhswish/gen/f32-vhswish-scalar-u4.c
src/f32-vlog/gen/f32-vlog-scalar-log.c
src/f32-vlog/gen/f32-vlog-scalar-rational-3-3-div.c
src/f32-vlrelu/gen/f32-vlrelu-scalar-u1.c
src/f32-vlrelu/gen/f32-vlrelu-scalar-u2.c
src/f32-vlrelu/gen/f32-vlrelu-scalar-u4.c
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/sse2_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,6 +68,7 @@ SET(ALL_SSE2_MICROKERNEL_SRCS
src/f32-velu/gen/f32-velu-sse2-rr2-p6-u16.c
src/f32-velu/gen/f32-velu-sse2-rr2-p6-u20.c
src/f32-velu/gen/f32-velu-sse2-rr2-p6-u24.c
src/f32-vlog/gen/f32-vlog-sse2-rational-3-3-div.c
src/f32-vlrelu/gen/f32-vlrelu-sse2-u4.c
src/f32-vlrelu/gen/f32-vlrelu-sse2-u8.c
src/f32-vrnd/gen/f32-vrndd-sse2-u4.c
Expand Down
1 change: 1 addition & 0 deletions cmake/gen/wasmsimd_microkernels.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -803,6 +803,7 @@ SET(ALL_WASMSIMD_MICROKERNEL_SRCS
src/f32-vhswish/gen/f32-vhswish-wasmsimd-u4.c
src/f32-vhswish/gen/f32-vhswish-wasmsimd-u8.c
src/f32-vhswish/gen/f32-vhswish-wasmsimd-u16.c
src/f32-vlog/gen/f32-vlog-wasmsimd-rational-3-3-div.c
src/f32-vlrelu/gen/f32-vlrelu-wasmsimd-iminmax-u4.c
src/f32-vlrelu/gen/f32-vlrelu-wasmsimd-iminmax-u8.c
src/f32-vlrelu/gen/f32-vlrelu-wasmsimd-laneselect-u4.c
Expand Down
1 change: 1 addition & 0 deletions gen/avx2_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -226,6 +226,7 @@ ALL_AVX2_MICROKERNEL_SRCS = [
"src/f32-velu/gen/f32-velu-avx2-rr1-p6-u64.c",
"src/f32-velu/gen/f32-velu-avx2-rr1-p6-u72.c",
"src/f32-velu/gen/f32-velu-avx2-rr1-p6-u80.c",
"src/f32-vlog/gen/f32-vlog-avx2-rational-3-3-div.c",
"src/f32-vscaleexpminusmax/gen/f32-vscaleexpminusmax-avx2-p5-u8.c",
"src/f32-vscaleexpminusmax/gen/f32-vscaleexpminusmax-avx2-p5-u16.c",
"src/f32-vscaleexpminusmax/gen/f32-vscaleexpminusmax-avx2-p5-u24.c",
Expand Down
1 change: 1 addition & 0 deletions gen/avx512f_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,6 +167,7 @@ ALL_AVX512F_MICROKERNEL_SRCS = [
"src/f32-velu/gen/f32-velu-avx512f-rr1-p6-u128.c",
"src/f32-vhswish/gen/f32-vhswish-avx512f-u16.c",
"src/f32-vhswish/gen/f32-vhswish-avx512f-u32.c",
"src/f32-vlog/gen/f32-vlog-avx512f-rational-3-3-div.c",
"src/f32-vlrelu/gen/f32-vlrelu-avx512f-u16.c",
"src/f32-vlrelu/gen/f32-vlrelu-avx512f-u32.c",
"src/f32-vrelu/gen/f32-vrelu-avx512f-u16.c",
Expand Down
1 change: 1 addition & 0 deletions gen/fma3_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -178,6 +178,7 @@ ALL_FMA3_MICROKERNEL_SRCS = [
"src/f32-qc8w-gemm/gen/f32-qc8w-gemm-8x16-minmax-fma3-broadcast.c",
"src/f32-vhswish/gen/f32-vhswish-fma3-u8.c",
"src/f32-vhswish/gen/f32-vhswish-fma3-u16.c",
"src/f32-vlog/gen/f32-vlog-fma3-rational-3-3-div.c",
"src/f32-vrsqrt/gen/f32-vrsqrt-fma3-rsqrt-u8.c",
"src/f32-vrsqrt/gen/f32-vrsqrt-fma3-rsqrt-u16.c",
"src/f32-vrsqrt/gen/f32-vrsqrt-fma3-rsqrt-u32.c",
Expand Down
1 change: 1 addition & 0 deletions gen/neon_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -340,6 +340,7 @@ ALL_NEON_MICROKERNEL_SRCS = [
"src/f32-vhswish/gen/f32-vhswish-neon-u4.c",
"src/f32-vhswish/gen/f32-vhswish-neon-u8.c",
"src/f32-vhswish/gen/f32-vhswish-neon-u16.c",
"src/f32-vlog/gen/f32-vlog-neon-rational-3-3-div.c",
"src/f32-vlrelu/gen/f32-vlrelu-neon-u4.c",
"src/f32-vlrelu/gen/f32-vlrelu-neon-u8.c",
"src/f32-vmulcaddc/gen/f32-vmulcaddc-c4-minmax-neon-2x.c",
Expand Down
1 change: 1 addition & 0 deletions gen/scalar_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -454,6 +454,7 @@ ALL_SCALAR_MICROKERNEL_SRCS = [
"src/f32-vhswish/gen/f32-vhswish-scalar-u2.c",
"src/f32-vhswish/gen/f32-vhswish-scalar-u4.c",
"src/f32-vlog/gen/f32-vlog-scalar-log.c",
"src/f32-vlog/gen/f32-vlog-scalar-rational-3-3-div.c",
"src/f32-vlrelu/gen/f32-vlrelu-scalar-u1.c",
"src/f32-vlrelu/gen/f32-vlrelu-scalar-u2.c",
"src/f32-vlrelu/gen/f32-vlrelu-scalar-u4.c",
Expand Down
1 change: 1 addition & 0 deletions gen/sse2_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,6 +64,7 @@ ALL_SSE2_MICROKERNEL_SRCS = [
"src/f32-velu/gen/f32-velu-sse2-rr2-p6-u16.c",
"src/f32-velu/gen/f32-velu-sse2-rr2-p6-u20.c",
"src/f32-velu/gen/f32-velu-sse2-rr2-p6-u24.c",
"src/f32-vlog/gen/f32-vlog-sse2-rational-3-3-div.c",
"src/f32-vlrelu/gen/f32-vlrelu-sse2-u4.c",
"src/f32-vlrelu/gen/f32-vlrelu-sse2-u8.c",
"src/f32-vrnd/gen/f32-vrndd-sse2-u4.c",
Expand Down
1 change: 1 addition & 0 deletions gen/wasmsimd_microkernels.bzl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -799,6 +799,7 @@ ALL_WASMSIMD_MICROKERNEL_SRCS = [
"src/f32-vhswish/gen/f32-vhswish-wasmsimd-u4.c",
"src/f32-vhswish/gen/f32-vhswish-wasmsimd-u8.c",
"src/f32-vhswish/gen/f32-vhswish-wasmsimd-u16.c",
"src/f32-vlog/gen/f32-vlog-wasmsimd-rational-3-3-div.c",
"src/f32-vlrelu/gen/f32-vlrelu-wasmsimd-iminmax-u4.c",
"src/f32-vlrelu/gen/f32-vlrelu-wasmsimd-iminmax-u8.c",
"src/f32-vlrelu/gen/f32-vlrelu-wasmsimd-laneselect-u4.c",
Expand Down
1 change: 1 addition & 0 deletions scripts/generate-benchmarks.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@ tools/generate-vunary-benchmark.py --spec test/f32-vsigmoid.yaml --output bench/
tools/generate-vunary-benchmark.py --spec test/f32-vsqr.yaml --output bench/f32-vsqr.cc &
tools/generate-vunary-benchmark.py --spec test/f32-vsqrt.yaml --output bench/f32-vsqrt.cc &
tools/generate-vunary-benchmark.py --spec test/f32-vtanh.yaml --output bench/f32-vtanh.cc &
tools/generate-vunary-benchmark.py --spec test/f32-vlog.yaml --output bench/f32-vlog.cc &

### Tests for VLRelu micro-kernels
tools/generate-vunary-benchmark.py --spec test/f16-vlrelu.yaml --output bench/f16-vlrelu.cc &
Expand Down
9 changes: 9 additions & 0 deletions scripts/generate-f32-vlog.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,4 +7,13 @@
#################################### Scalar ###################################
tools/xngen src/f32-vlog/scalar-log.c.in -D BATCH_TILES=1,2,4 -o src/f32-vlog/gen/f32-vlog-scalar-log.c &

##################################### SIMD #####################################
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=scalar -D BATCH_TILES=1,2,4,8 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-scalar-rational-3-3-div.c &
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=sse2 -D BATCH_TILES=4,8,12,16 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-sse2-rational-3-3-div.c &
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=avx2 -D BATCH_TILES=8,16,24,32 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-avx2-rational-3-3-div.c &
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=fma3 -D BATCH_TILES=8,16,24,32 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-fma3-rational-3-3-div.c &
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=avx512f -D BATCH_TILES=16,32,48,64 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-avx512f-rational-3-3-div.c &
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=neon -D BATCH_TILES=4,8,12,16 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-neon-rational-3-3-div.c &
tools/xngen src/f32-vlog/rational-3-3.c.in -D ARCH=wasmsimd -D BATCH_TILES=4,8,12,16 -D DIV=DIV -o src/f32-vlog/gen/f32-vlog-wasmsimd-rational-3-3-div.c &

wait
Loading