diff --git a/paddle/fluid/platform/transform_test.cu b/paddle/fluid/platform/transform_test.cu index 7b5cfd8f43473dc6bc784e98bd26fdd9e0ba9994..f65d1f60100edc85ba9745ed36f26a0ed160d80f 100644 --- a/paddle/fluid/platform/transform_test.cu +++ b/paddle/fluid/platform/transform_test.cu @@ -18,11 +18,12 @@ limitations under the License. */ #include "paddle/fluid/platform/hostdevice.h" #include "paddle/fluid/platform/transform.h" +namespace { + template class Scale { public: explicit Scale(const T& scale) : scale_(scale) {} - HOSTDEVICE T operator()(const T& a) const { return a * scale_; } private: @@ -35,11 +36,23 @@ class Multiply { HOSTDEVICE T operator()(const T& a, const T& b) const { return a * b; } }; +} // namespace + +using paddle::memory::Alloc; +using paddle::memory::Free; +using paddle::memory::Copy; + +using paddle::platform::CPUPlace; +using paddle::platform::CUDAPlace; +using paddle::platform::CPUDeviceContext; +using paddle::platform::CUDADeviceContext; + +using paddle::platform::Transform; + TEST(Transform, CPUUnary) { - using namespace paddle::platform; CPUDeviceContext ctx; float buf[4] = {0.1, 0.2, 0.3, 0.4}; - Transform trans; + Transform trans; trans(ctx, buf, buf + 4, buf, Scale(10)); for (int i = 0; i < 4; ++i) { ASSERT_NEAR(buf[i], static_cast(i + 1), 1e-5); @@ -47,14 +60,12 @@ TEST(Transform, CPUUnary) { } TEST(Transform, GPUUnary) { - using namespace paddle::platform; - using namespace paddle::memory; CUDAPlace gpu0(0); CUDADeviceContext ctx(gpu0); float cpu_buf[4] = {0.1, 0.2, 0.3, 0.4}; float* gpu_buf = static_cast(Alloc(gpu0, sizeof(float) * 4)); Copy(gpu0, gpu_buf, CPUPlace(), cpu_buf, sizeof(cpu_buf), ctx.stream()); - Transform trans; + Transform trans; trans(ctx, gpu_buf, gpu_buf + 4, gpu_buf, Scale(10)); ctx.Wait(); Copy(CPUPlace(), cpu_buf, gpu0, gpu_buf, sizeof(cpu_buf), ctx.stream()); @@ -65,10 +76,8 @@ TEST(Transform, GPUUnary) { } TEST(Transform, CPUBinary) { - using namespace paddle::platform; - using namespace paddle::memory; int buf[4] = {1, 2, 3, 4}; - Transform trans; + Transform trans; CPUDeviceContext ctx; trans(ctx, buf, buf + 4, buf, buf, Multiply()); for (int i = 0; i < 4; ++i) { @@ -77,14 +86,12 @@ TEST(Transform, CPUBinary) { } TEST(Transform, GPUBinary) { - using namespace paddle::platform; - using namespace paddle::memory; int buf[4] = {1, 2, 3, 4}; CUDAPlace gpu0(0); CUDADeviceContext ctx(gpu0); int* gpu_buf = static_cast(Alloc(gpu0, sizeof(buf))); Copy(gpu0, gpu_buf, CPUPlace(), buf, sizeof(buf), ctx.stream()); - Transform trans; + Transform trans; trans(ctx, gpu_buf, gpu_buf + 4, gpu_buf, gpu_buf, Multiply()); ctx.Wait(); Copy(CPUPlace(), buf, gpu0, gpu_buf, sizeof(buf), ctx.stream());