Hash :
3402d523
Author :
Date :
2018-10-30T15:14:52
Try to reduce variance in angle_perftests. This change does a few things: - make perf test runner script print % variation instead of stddev This makes it a bit more clear how much variance there is. - stabilize CPU in the render perf tests Setting a thread affinity and priority should stop from switching cores during the run. Hopefully can prevent background noise from changing the test results. - warm up the benchmark with a few iterations This should hopefully make the test results a bit more stable. - output a new normalized perf result value The new result is normalized against the number of iterations. So it should hopefully be stable even if the number of iterations is changed. - increases the iteration count in the draw call perf tests. These tests were completely dominated by SwapBuffers time. Increasing the iterations per step means we actually are bottlenecked on CPU time instead. Bug: angleproject:2923 Change-Id: I5ee347cf93df239ac33b83dc5effe4c21e066736 Reviewed-on: https://chromium-review.googlesource.com/c/1303679 Commit-Queue: Jamie Madill <jmadill@chromium.org> Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org> Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176
//
// Copyright (c) 2018 The ANGLE Project Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
//
// DispatchComputePerf:
// Performance tests for ANGLE DispatchCompute call overhead.
//
#include "ANGLEPerfTest.h"
#include "shader_utils.h"
namespace
{
unsigned int kIterationsPerStep = 50;
struct DispatchComputePerfParams final : public RenderTestParams
{
DispatchComputePerfParams()
{
iterationsPerStep = kIterationsPerStep;
majorVersion = 3;
minorVersion = 1;
}
std::string suffix() const override;
unsigned int localSizeX = 16;
unsigned int localSizeY = 16;
unsigned int textureWidth = 32;
unsigned int textureHeight = 32;
};
std::string DispatchComputePerfParams::suffix() const
{
std::stringstream suffixStr;
suffixStr << RenderTestParams::suffix();
if (eglParameters.deviceType == EGL_PLATFORM_ANGLE_DEVICE_TYPE_NULL_ANGLE)
{
suffixStr << "_null";
}
return suffixStr.str();
}
std::ostream &operator<<(std::ostream &os, const DispatchComputePerfParams ¶ms)
{
os << params.suffix().substr(1);
return os;
}
class DispatchComputePerfBenchmark : public ANGLERenderTest,
public ::testing::WithParamInterface<DispatchComputePerfParams>
{
public:
DispatchComputePerfBenchmark();
void initializeBenchmark() override;
void destroyBenchmark() override;
void drawBenchmark() override;
private:
void initComputeShader();
void initTextures();
GLuint mProgram = 0;
GLuint mReadTexture = 0;
GLuint mWriteTexture = 0;
GLuint mDispatchX = 0;
GLuint mDispatchY = 0;
};
DispatchComputePerfBenchmark::DispatchComputePerfBenchmark()
: ANGLERenderTest("DispatchComputePerf", GetParam())
{
}
void DispatchComputePerfBenchmark::initializeBenchmark()
{
const auto ¶ms = GetParam();
initComputeShader();
initTextures();
glUseProgram(mProgram);
glActiveTexture(GL_TEXTURE0);
glBindTexture(GL_TEXTURE_2D, mReadTexture);
glUniform1i(glGetUniformLocation(mProgram, "readTexture"), 0);
glBindImageTexture(4, mWriteTexture, 0, GL_FALSE, 0, GL_WRITE_ONLY, GL_R32F);
mDispatchX = params.textureWidth / params.localSizeX;
mDispatchY = params.textureHeight / params.localSizeY;
ASSERT_GL_NO_ERROR();
}
void DispatchComputePerfBenchmark::initComputeShader()
{
const std::string &csSource =
R"(#version 310 es
#define LOCAL_SIZE_X 16
#define LOCAL_SIZE_Y 16
layout(local_size_x=LOCAL_SIZE_X, local_size_y=LOCAL_SIZE_Y) in;
precision highp float;
uniform sampler2D readTexture;
layout(r32f, binding = 4) writeonly uniform highp image2D outImage;
void main() {
float sum = 0.;
sum += texelFetch(readTexture, ivec2(gl_GlobalInvocationID.xy), 0).r;
imageStore(outImage, ivec2(gl_GlobalInvocationID.xy), vec4(sum));
})";
mProgram = CompileComputeProgram(csSource, false);
ASSERT_NE(0u, mProgram);
}
void DispatchComputePerfBenchmark::initTextures()
{
const auto ¶ms = GetParam();
unsigned int textureDataSize = params.textureWidth * params.textureHeight;
std::vector<GLfloat> textureInputData(textureDataSize, 0.2f);
std::vector<GLfloat> textureOutputData(textureDataSize, 0.1f);
glGenTextures(1, &mReadTexture);
glBindTexture(GL_TEXTURE_2D, mReadTexture);
glTexImage2D(GL_TEXTURE_2D, 0, GL_R32F, params.textureWidth, params.textureHeight, 0, GL_RED,
GL_FLOAT, textureInputData.data());
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glGenTextures(1, &mWriteTexture);
glBindTexture(GL_TEXTURE_2D, mWriteTexture);
glTexStorage2D(GL_TEXTURE_2D, 1, GL_R32F, params.textureWidth, params.textureHeight);
glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, params.textureWidth, params.textureHeight, GL_RED,
GL_FLOAT, textureOutputData.data());
ASSERT_GL_NO_ERROR();
}
void DispatchComputePerfBenchmark::destroyBenchmark()
{
glDeleteProgram(mProgram);
glDeleteTextures(1, &mReadTexture);
glDeleteTextures(1, &mWriteTexture);
}
void DispatchComputePerfBenchmark::drawBenchmark()
{
const auto ¶ms = GetParam();
for (unsigned int it = 0; it < params.iterationsPerStep; it++)
{
glDispatchCompute(mDispatchX, mDispatchY, 1);
glMemoryBarrier(GL_TEXTURE_UPDATE_BARRIER_BIT);
}
ASSERT_GL_NO_ERROR();
}
DispatchComputePerfParams DispatchComputePerfOpenGLOrGLESParams(bool useNullDevice)
{
DispatchComputePerfParams params;
params.eglParameters = angle::egl_platform::OPENGL_OR_GLES(useNullDevice);
return params;
}
TEST_P(DispatchComputePerfBenchmark, Run)
{
run();
}
ANGLE_INSTANTIATE_TEST(DispatchComputePerfBenchmark,
DispatchComputePerfOpenGLOrGLESParams(true),
DispatchComputePerfOpenGLOrGLESParams(false));
} // namespace