a2d25ec0ef
Also split bench into run/compile variants to measure the effect: Before …f16_compile 1x …f16_run 1.02x …srgb_compile 1.56x …srgb_run 1.61x After …f16_run 1x …f16_compile 1.01x …srgb_compile 1.58x …srgb_run 1.59x CQ_INCLUDE_TRYBOTS=skia.primary:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD Change-Id: I8e65fb2acdbb05ccc0b3894f16d7646603c3e74d Reviewed-on: https://skia-review.googlesource.com/6621 Reviewed-by: Herb Derby <herb@google.com> Commit-Queue: Mike Klein <mtklein@chromium.org>
78 lines
2.4 KiB
C++
78 lines
2.4 KiB
C++
/*
|
|
* Copyright 2016 Google Inc.
|
|
*
|
|
* Use of this source code is governed by a BSD-style license that can be
|
|
* found in the LICENSE file.
|
|
*/
|
|
|
|
#include "Benchmark.h"
|
|
#include "SkOpts.h"
|
|
#include "SkRasterPipeline.h"
|
|
|
|
static const int N = 1023;
|
|
|
|
static uint64_t dst[N]; // sRGB or F16
|
|
static uint32_t src[N]; // sRGB
|
|
static uint8_t mask[N]; // 8-bit linear
|
|
|
|
// We'll build up a somewhat realistic useful pipeline:
|
|
// - load srgb src
|
|
// - scale src by 8-bit mask
|
|
// - load srgb/f16 dst
|
|
// - src = srcover(dst, src)
|
|
// - store src back as srgb/f16
|
|
|
|
template <bool kF16, bool kCompiled>
|
|
class SkRasterPipelineBench : public Benchmark {
|
|
public:
|
|
bool isSuitableFor(Backend backend) override { return backend == kNonRendering_Backend; }
|
|
const char* onGetName() override {
|
|
switch ((int)kCompiled << 1 | (int)kF16) {
|
|
case 0: return "SkRasterPipeline_srgb_run";
|
|
case 1: return "SkRasterPipeline_f16_run";
|
|
case 2: return "SkRasterPipeline_srgb_compile";
|
|
case 3: return "SkRasterPipeline_f16_compile";
|
|
}
|
|
return "whoops";
|
|
}
|
|
|
|
void onDraw(int loops, SkCanvas*) override {
|
|
void* mask_ctx = mask;
|
|
void* src_ctx = src;
|
|
void* dst_ctx = dst;
|
|
|
|
SkRasterPipeline p;
|
|
p.append(SkRasterPipeline::load_8888, &src_ctx);
|
|
p.append_from_srgb(kUnpremul_SkAlphaType);
|
|
p.append(SkRasterPipeline::scale_u8, &mask_ctx);
|
|
if (kF16) {
|
|
p.append(SkRasterPipeline::load_f16_d, &dst_ctx);
|
|
} else {
|
|
p.append(SkRasterPipeline::load_8888_d, &dst_ctx);
|
|
p.append_from_srgb_d(kPremul_SkAlphaType);
|
|
}
|
|
p.append(SkRasterPipeline::srcover);
|
|
if (kF16) {
|
|
p.append(SkRasterPipeline::store_f16, &dst_ctx);
|
|
} else {
|
|
p.append(SkRasterPipeline::to_srgb);
|
|
p.append(SkRasterPipeline::store_8888, &dst_ctx);
|
|
}
|
|
|
|
if (kCompiled) {
|
|
auto compiled = p.compile();
|
|
while (loops --> 0) {
|
|
compiled(0,0, N);
|
|
}
|
|
} else {
|
|
while (loops --> 0) {
|
|
p.run(0,0, N);
|
|
}
|
|
}
|
|
}
|
|
};
|
|
DEF_BENCH( return (new SkRasterPipelineBench< true, true>); )
|
|
DEF_BENCH( return (new SkRasterPipelineBench<false, true>); )
|
|
DEF_BENCH( return (new SkRasterPipelineBench< true, false>); )
|
|
DEF_BENCH( return (new SkRasterPipelineBench<false, false>); )
|