Home
last modified time | relevance | path

Searched refs:warpSize (Results 1 – 11 of 11) sorted by relevance

/external/clang/test/SemaCUDA/
Dcuda-builtin-vars.cu35 out[i++] = warpSize; in kernel()
36 …warpSize = 0; // expected-error {{cannot assign to variable 'warpSize' with const-qualified type '… in kernel()
56 const void *wsptr = &warpSize; in kernel()
/external/clang/lib/Headers/
D__clang_cuda_intrinsics.h39 int __width = warpSize) { \
41 ((warpSize - __width) << 8) | (__Mask)); \
44 int __width = warpSize) { \
46 ((warpSize - __width) << 8) | (__Mask)); \
49 int __width = warpSize) { \
54 int __width = warpSize) { \
69 unsigned long long __in, int __offset, int __width = warpSize) { \
74 int __width = warpSize) { \
Dcuda_builtin_vars.h120 __attribute__((device)) const int warpSize = 32; variable
/external/eigen/unsupported/Eigen/CXX11/src/Tensor/
DTensorReductionCuda.h170 for (int offset = warpSize/2; offset > 0; offset /= 2) { in FullReductionKernel()
171 reducer.reduce(__shfl_down(accum, offset, warpSize), &accum); in FullReductionKernel()
174 if ((threadIdx.x & (warpSize - 1)) == 0) { in FullReductionKernel()
246 for (int offset = warpSize/2; offset > 0; offset /= 2) { in FullReductionKernelHalfFloat()
247 reducer.reducePacket(__shfl_down(accum, offset, warpSize), &accum); in FullReductionKernelHalfFloat()
250 if ((threadIdx.x & (warpSize - 1)) == 0) { in FullReductionKernelHalfFloat()
428 for (int offset = warpSize/2; offset > 0; offset /= 2) {
432 if ((threadIdx.x & (warpSize - 1)) == 0) {
518 for (int offset = warpSize/2; offset > 0; offset /= 2) {
519 reducer.reducePacket(__shfl_down(reduced_val1, offset, warpSize), &reduced_val1);
[all …]
DTensorConvolution.h858 const int warpSize = 32;
883 const int inner_dim = maxSharedMem / ((warpSize + kernel_size) * sizeof(Scalar));
887 block_size.x = numext::mini(warpSize, maxX);
/external/opencv/cvaux/src/
Dcvcalibfilter.cpp750 …stereo.warpSize = cvSize( cvRound(cameraParams[0].imgSize[0]), cvRound(cameraParams[0].imgSize[1])… in LoadCameraParams()
807 … rectMap[i][0] = cvCreateMat(stereo.warpSize.height,stereo.warpSize.width,CV_32FC1); in Rectify()
808 … rectMap[i][1] = cvCreateMat(stereo.warpSize.height,stereo.warpSize.width,CV_32FC1); in Rectify()
Dcvepilines.cpp326 stereoCamera->warpSize.height, in icvComputeCoeffForStereo()
1302 CvSize* warpSize, in icvGetQuadsTransform() argument
1656 warpSize->width = (int)warpWidth; in icvGetQuadsTransform()
1657 warpSize->height = (int)warpHeight; in icvGetQuadsTransform()
1659 warpSize->width = cvRound(warpWidth-1); in icvGetQuadsTransform()
1660 warpSize->height = cvRound(warpHeight-1); in icvGetQuadsTransform()
1663 warpSize->width = imageSize.width; in icvGetQuadsTransform()
1664 warpSize->height = imageSize.height; in icvGetQuadsTransform()
1677 CvSize* warpSize, in icvGetQuadsTransformNew() argument
1725 warpSize, in icvGetQuadsTransformNew()
[all …]
/external/clang/test/CodeGenCUDA/
Dcuda-builtin-vars.cu25 out[i++] = warpSize; // CHECK: store i32 32, in kernel()
/external/eigen/test/
Dcuda_common.h93 std::cout << " warpSize: " << deviceProp.warpSize << "\n"; in ei_test_init_cuda()
/external/opencv/cvaux/include/
Dcvaux.h532 CvSize warpSize; member
693 CvSize* warpSize,
/external/eigen/Eigen/src/Core/arch/CUDA/
DHalf.h545 …e__ EIGEN_STRONG_INLINE Eigen::half __shfl_xor(Eigen::half var, int laneMask, int width=warpSize) {