generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

switched to Input/Output Array in gpu::gemm

Browse Source
This commit is contained in:
Vladislav Vinogradov
2013-04-26 12:22:56 +04:00
parent 8fcef225fb
commit 948661d722
2 changed files with 23 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
modules/gpuarithm/include/opencv2/gpuarithm.hpp
View File

@@ -350,9 +350,8 @@ static inline void sqrIntegral(InputArray src, OutputArray sqsum, Stream& stream
sqrIntegral(src, sqsum, buffer, stream);
}
//! implements generalized matrix product algorithm GEMM from BLAS
CV_EXPORTS void gemm(const GpuMat& src1, const GpuMat& src2, double alpha,
const GpuMat& src3, double beta, GpuMat& dst, int flags = 0, Stream& stream = Stream::Null());
CV_EXPORTS void gemm(InputArray src1, InputArray src2, double alpha,
InputArray src3, double beta, OutputArray dst, int flags = 0, Stream& stream = Stream::Null());
//! performs per-element multiplication of two full (not packed) Fourier spectrums
//! supports 32FC2 matrixes only (interleaved format)

21
modules/gpuarithm/src/arithm.cpp
View File

@@ -47,7 +47,7 @@ using namespace cv::gpu;
#if !defined (HAVE_CUDA) || defined (CUDA_DISABLER)
void cv::gpu::gemm(const GpuMat&, const GpuMat&, double, const GpuMat&, double, GpuMat&, int, Stream&) { throw_no_cuda(); }
void cv::gpu::gemm(InputArray, InputArray, double, InputArray, double, OutputArray, int, Stream&) { throw_no_cuda(); }
void cv::gpu::mulSpectrums(const GpuMat&, const GpuMat&, GpuMat&, int, bool, Stream&) { throw_no_cuda(); }
void cv::gpu::mulAndScaleSpectrums(const GpuMat&, const GpuMat&, GpuMat&, int, float, bool, Stream&) { throw_no_cuda(); }
@@ -164,21 +164,25 @@ namespace
////////////////////////////////////////////////////////////////////////
// gemm
void cv::gpu::gemm(const GpuMat& src1, const GpuMat& src2, double alpha, const GpuMat& src3, double beta, GpuMat& dst, int flags, Stream& stream)
void cv::gpu::gemm(InputArray _src1, InputArray _src2, double alpha, InputArray _src3, double beta, OutputArray _dst, int flags, Stream& stream)
{
#ifndef HAVE_CUBLAS
(void)src1;
(void)src2;
(void) _src1;
(void) _src2;
(void) alpha;
(void)src3;
(void) _src3;
(void) beta;
(void)dst;
(void) _dst;
(void) flags;
(void) stream;
CV_Error(cv::Error::StsNotImplemented, "The library was build without CUBLAS");
CV_Error(:Error::StsNotImplemented, "The library was build without CUBLAS");
#else
// CUBLAS works with column-major matrices
GpuMat src1 = _src1.getGpuMat();
GpuMat src2 = _src2.getGpuMat();
GpuMat src3 = _src3.getGpuMat();
CV_Assert( src1.type() == CV_32FC1 || src1.type() == CV_32FC2 || src1.type() == CV_64FC1 || src1.type() == CV_64FC2 );
CV_Assert( src2.type() == src1.type() && (src3.empty() || src3.type() == src1.type()) );
@@ -206,7 +210,8 @@ void cv::gpu::gemm(const GpuMat& src1, const GpuMat& src2, double alpha, const G
CV_Assert( src1Size.width == src2Size.height );
CV_Assert( src3.empty() || src3Size == dstSize );
dst.create(dstSize, src1.type());
_dst.create(dstSize, src1.type());
GpuMat dst = _dst.getGpuMat();
if (beta != 0)
{