Skip to content

P
PaddleDetection

PaddlePaddle / PaddleDetection
大约 1 年 前同步成功

通知 695
Star 11112
Fork 2696
P
PaddleDetection
提交 248149f4 编写于 作者: X xzl
浏览文件
操作

add depthwiseconv test and fix the little bug of the convOpTest

上级 5b07d4e0
隐藏空白更改
内联 并排
Showing with 202 addition and 147 deletion
paddle/function/ConvOpTest.cpp
浏览文件 @ 248149f4
......@@ -38,76 +38,76 @@ public:
for (size_t filterSize : {1, 3, 5}) {
for (size_t inputChannels : {3, 64}) {
for (size_t outputChannels : {3, 64}) {
for (size_t groups : {1, 3, 64}) {
if (inputChannels > outputChannels) break;
if (groups != 1 &&
(inputChannels != groups || outputChannels % groups != 0))
continue;
if (!useGroups) groups = 1;
for (size_t stride : {1, 2}) {
for (size_t padding : {0, 1}) {
if (padding >= filterSize) break;
size_t outputSize =
(inputSize - filterSize + 2 * padding + stride) /
stride;
VLOG(3) << " batchSize=" << batchSize
<< " inputChannels=" << inputChannels
<< " inputHeight=" << inputSize
<< " inputWidth=" << inputSize
<< " outputChannels=" << outputChannels
<< " filterHeight=" << filterSize
<< " filterWidth=" << filterSize
<< " outputHeight=" << outputSize
<< " outputWidth=" << outputSize
<< " stride=" << stride << " padding=" << padding;
std::vector<size_t> paddings = {padding, padding};
std::vector<size_t> strides = {stride, stride};
Compare2Function<DType1, DType2> test(
conv1,
conv2,
FuncConfig()
.set("paddings", paddings)
.set("strides", strides)
.set("groups", groups)
.set("algo", algo));
TensorShape input{
batchSize, inputChannels, inputSize, inputSize};
TensorShape filter;
if (groups > 1)
filter = TensorShape({groups,
outputChannels / groups,
inputChannels / groups,
filterSize,
filterSize});
else
filter = TensorShape({outputChannels,
inputChannels,
filterSize,
filterSize});
TensorShape output{
batchSize, outputChannels, outputSize, outputSize};
if (type == kForwardTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.run();
} else if (type == kBackwardInputTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, input),
ADD_TO);
test.run();
} else if (type == kBackwardFilterTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.run();
}
if (inputChannels > outputChannels) break;
size_t groups;
if (!useGroups) {
groups = 1;
} else {
if (outputChannels % inputChannels != 0) continue;
groups = inputChannels;
}
for (size_t stride : {1, 2}) {
for (size_t padding : {0, 1}) {
if (padding >= filterSize) break;
size_t outputSize =
(inputSize - filterSize + 2 * padding + stride) / stride;
VLOG(3) << " batchSize=" << batchSize
<< " inputChannels=" << inputChannels
<< " inputHeight=" << inputSize
<< " inputWidth=" << inputSize
<< " outputChannels=" << outputChannels
<< " filterHeight=" << filterSize
<< " filterWidth=" << filterSize
<< " outputHeight=" << outputSize
<< " outputWidth=" << outputSize
<< " stride=" << stride << " padding=" << padding;
std::vector<size_t> paddings = {padding, padding};
std::vector<size_t> strides = {stride, stride};
Compare2Function<DType1, DType2> test(
conv1,
conv2,
FuncConfig()
.set("paddings", paddings)
.set("strides", strides)
.set("groups", groups)
.set("algo", algo));
TensorShape input{
batchSize, inputChannels, inputSize, inputSize};
TensorShape filter;
if (groups > 1)
filter = TensorShape({groups,
outputChannels / groups,
inputChannels / groups,
filterSize,
filterSize});
else
filter = TensorShape({outputChannels,
inputChannels,
filterSize,
filterSize});
TensorShape output{
batchSize, outputChannels, outputSize, outputSize};
if (type == kForwardTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.run();
} else if (type == kBackwardInputTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, input), ADD_TO);
test.run();
} else if (type == kBackwardFilterTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, filter),
ADD_TO);
test.run();
}
}
}
......@@ -136,77 +136,78 @@ public:
for (size_t filterWidth : {3, 7}) {
for (size_t inputChannels : {7}) {
for (size_t outputChannels : {7}) {
for (size_t groups : {1, 7}) {
if (groups != 1 && (inputChannels != groups ||
outputChannels % groups != 0))
continue;
if (!useGroups) groups = 1;
size_t stride = 1;
size_t padding = 0;
size_t outputHeight =
(inputHeight - filterHeight + 2 * padding + stride) /
stride;
size_t outputWidth =
(inputWidth - filterWidth + 2 * padding + stride) /
stride;
VLOG(3) << " batchSize=" << batchSize
<< " inputChannels=" << inputChannels
<< " inputHeight=" << inputHeight
<< " inputWidth=" << inputWidth
<< " outputChannels=" << outputChannels
<< " filterHeight=" << filterHeight
<< " filterWidth=" << filterWidth
<< " outputHeight=" << outputHeight
<< " outputWidth=" << outputWidth
<< " stride=" << stride << " padding=" << padding;
std::vector<size_t> paddings = {padding, padding};
std::vector<size_t> strides = {stride, stride};
Compare2Function<DType1, DType2> test(
conv1,
conv2,
FuncConfig()
.set("paddings", paddings)
.set("strides", strides)
.set("groups", groups)
.set("algo", algo));
TensorShape input{
batchSize, inputChannels, inputHeight, inputWidth};
TensorShape filter;
if (groups > 1)
filter = TensorShape({groups,
outputChannels / groups,
inputChannels / groups,
filterHeight,
filterWidth});
else
filter = TensorShape({outputChannels,
inputChannels,
filterHeight,
filterWidth});
TensorShape output{
batchSize, outputChannels, outputHeight, outputWidth};
if (type == kForwardTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.run();
} else if (type == kBackwardInputTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, input),
ADD_TO);
test.run();
} else if (type == kBackwardFilterTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.run();
}
size_t groups;
if (!useGroups) {
groups = 1;
} else {
if (outputChannels % inputChannels != 0) continue;
groups = inputChannels;
}
size_t stride = 1;
size_t padding = 0;
size_t outputHeight =
(inputHeight - filterHeight + 2 * padding + stride) /
stride;
size_t outputWidth =
(inputWidth - filterWidth + 2 * padding + stride) /
stride;
VLOG(3) << " batchSize=" << batchSize
<< " inputChannels=" << inputChannels
<< " inputHeight=" << inputHeight
<< " inputWidth=" << inputWidth
<< " outputChannels=" << outputChannels
<< " filterHeight=" << filterHeight
<< " filterWidth=" << filterWidth
<< " outputHeight=" << outputHeight
<< " outputWidth=" << outputWidth
<< " stride=" << stride << " padding=" << padding;
std::vector<size_t> paddings = {padding, padding};
std::vector<size_t> strides = {stride, stride};
Compare2Function<DType1, DType2> test(
conv1,
conv2,
FuncConfig()
.set("paddings", paddings)
.set("strides", strides)
.set("groups", groups)
.set("algo", algo));
TensorShape input{
batchSize, inputChannels, inputHeight, inputWidth};
TensorShape filter;
if (groups > 1)
filter = TensorShape({groups,
outputChannels / groups,
inputChannels / groups,
filterHeight,
filterWidth});
else
filter = TensorShape({outputChannels,
inputChannels,
filterHeight,
filterWidth});
TensorShape output{
batchSize, outputChannels, outputHeight, outputWidth};
if (type == kForwardTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.run();
} else if (type == kBackwardInputTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, filter));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, input), ADD_TO);
test.run();
} else if (type == kBackwardFilterTest) {
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, output));
test.addInputs(BufferArg(VALUE_TYPE_FLOAT, input));
test.addOutputs(BufferArg(VALUE_TYPE_FLOAT, filter),
ADD_TO);
test.run();
}
}
}
......@@ -218,6 +219,8 @@ public:
}
};
// ======Start Convolution TEST======
TEST(Forward, GEMM) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_CPU> test(
"NaiveConv-CPU", "GemmConv-CPU", kForwardTest, false);
......@@ -228,24 +231,76 @@ TEST(Forward, GEMM) {
#ifndef PADDLE_ONLY_CPU
TEST(Forward, GEMM2) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test(
"GemmConv-CPU", "GemmConv-GPU", kForwardTest);
"GemmConv-CPU", "GemmConv-GPU", kForwardTest, false);
ConvolutionTest2<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test2(
"GemmConv-CPU", "GemmConv-GPU", kForwardTest);
"GemmConv-CPU", "GemmConv-GPU", kForwardTest, false);
}
TEST(BackwardInput, GEMM) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test(
"GemmConvGradInput-CPU", "GemmConvGradInput-GPU", kBackwardInputTest);
"GemmConvGradInput-CPU",
"GemmConvGradInput-GPU",
kBackwardInputTest,
false);
ConvolutionTest2<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test2(
"GemmConvGradInput-CPU", "GemmConvGradInput-GPU", kBackwardInputTest);
"GemmConvGradInput-CPU",
"GemmConvGradInput-GPU",
kBackwardInputTest,
false);
}
TEST(BackwardFilter, GEMM) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test(
"GemmConvGradFilter-CPU", "GemmConvGradFilter-GPU", kBackwardFilterTest);
"GemmConvGradFilter-CPU",
"GemmConvGradFilter-GPU",
kBackwardFilterTest,
false);
ConvolutionTest2<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test2(
"GemmConvGradFilter-CPU", "GemmConvGradFilter-GPU", kBackwardFilterTest);
"GemmConvGradFilter-CPU",
"GemmConvGradFilter-GPU",
kBackwardFilterTest,
false);
}
#endif
// ======End Convolution TEST======
// ======Start DepthwiseConvolution TEST======
// TODO(zhaolong) The depthwise convolution cpu test will be added when the cpu
// version of depthwiseConv is implemented.
#ifndef PADDLE_ONLY_CPU
TEST(DepthwiseConvForward, GEMM2) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test(
"GemmConv-CPU", "DepthwiseConv-GPU", kForwardTest);
ConvolutionTest2<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test2(
"GemmConv-CPU", "DepthwiseConv-GPU", kForwardTest);
}
TEST(DepthwiseConvBackwardInput, GEMM) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test(
"GemmConvGradInput-CPU",
"DepthwiseConvGradInput-GPU",
kBackwardInputTest);
ConvolutionTest2<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test2(
"GemmConvGradInput-CPU",
"DepthwiseConvGradInput-GPU",
kBackwardInputTest);
}
TEST(DepthwiseConvBackwardFilter, GEMM) {
ConvolutionTest<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test(
"GemmConvGradFilter-CPU",
"DepthwiseConvGradFilter-GPU",
kBackwardFilterTest);
ConvolutionTest2<DEVICE_TYPE_CPU, DEVICE_TYPE_GPU> test2(
"GemmConvGradFilter-CPU",
"DepthwiseConvGradFilter-GPU",
kBackwardFilterTest);
}
#endif
// ======End DepthwiseConvolution TEST======
} // namespace paddle
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或