| /external/ruy/ruy/ |
| D | pack_arm.cc | 37 int src_inc3, int src_rows, int src_zero_point, in Pack8bitColMajorForNeon() argument
207 [rows] "r"(src_rows), [src_zero_point] "r"(src_zero_point), in Pack8bitColMajorForNeon()
244 static_assert(offsetof(Params, src_rows) == RUY_OFFSET_SRC_ROWS, ""); in CheckOffsetsInPackParams8bit()
632 int src_inc3, int src_rows, in Pack8bitColMajorForNeonA55ish() argument
831 [ rows ] "r"(src_rows), in Pack8bitColMajorForNeonA55ish()
843 int src_inc3, int src_rows, int src_zero_point, std::int8_t* packed_ptr, in Pack8bitColMajorForNeonDotprodA55ish() argument
1078 [rows] "r"(src_rows), in Pack8bitColMajorForNeonDotprodA55ish()
1089 int src_inc3, int src_rows, in Pack8bitColMajorForNeonDotprod() argument
1578 [rows] "r"(src_rows), in Pack8bitColMajorForNeonDotprod()
1741 int src_inc3, int src_rows, float* packed_ptr) { in PackFloatColMajorForNeon() argument
[all …]
|
| D | pack_avx512.cc | 62 inline void ZeroHalf8bitAvx512(int src_rows, std::int8_t packed_zero_point,
72 const int non_trailing_blocks = (src_rows & ~31) >> 2;
104 int remaining_src_cols, int src_rows,
181 for (int k = 0; k < src_rows; k += 2 * kNumChunkedSrcRows) {
188 const int available_src_rows = src_rows - k - m * kNumChunkedSrcRows;
482 int src_rows, float* packed_ptr,
533 for (int k = 0; k < src_rows; k += 16) {
535 const int available_src_rows = src_rows - k - 8 * m;
621 inline void ZeroHalfFloatAvx512(int src_rows, float* packed_ptr) {
622 const int non_trailing_rows = src_rows & ~7;
[all …]
|
| D | pack_avx2_fma.cc | 69 int src_rows, std::int8_t* packed_ptr, std::int32_t* sums_ptr,
146 for (int k = 0; k < src_rows; k += kNumChunkedSrcRows) {
151 const int available_src_rows = src_rows - k;
547 int remaining_src_cols, int src_rows,
567 remaining_src_cols, src_rows, packed_ptr,
571 const bool trailing_data = (src_rows & kChunkedRowMask) > 0;
575 const int non_trailing_rows = src_rows & ~kChunkedRowMask;
577 const int dst_rows = (src_rows + 3) & ~3;
586 int src_rows, float* packed_ptr) {
595 src_ptr, zerobuf, src_stride, remaining_src_cols, src_rows, packed_ptr,
[all …]
|
| D | pack_avx.cc | 218 int src_rows, std::int8_t* packed_ptr,
296 for (int k = 0; k < src_rows; k += kNumChunkedSrcRows) {
301 const int available_src_rows = src_rows - k;
689 int remaining_src_cols, int src_rows,
709 remaining_src_cols, src_rows, packed_ptr,
713 const bool trailing_data = (src_rows & kChunkedRowMask) > 0;
717 const int non_trailing_rows = src_rows & ~kChunkedRowMask;
719 const int dst_rows = (src_rows + 3) & ~3;
728 int src_rows, float* packed_ptr) {
737 src_ptr, zerobuf, src_stride, remaining_src_cols, src_rows, packed_ptr,
[all …]
|
| D | pack_arm.h | 56 int src_rows, int src_cols, int block_row,
68 int src_inc3, int src_rows, int src_zero_point,
74 int src_inc3, int src_rows,
80 int src_inc3, int src_rows,
86 int src_inc3, int src_rows, int src_zero_point, std::int8_t* packed_ptr,
108 int src_rows;
118 int src_rows, int src_zero_point, int input_xor,
130 params->src_rows = src_rows;
345 int src_inc3, int src_rows, float* packed_ptr);
351 int src_rows, float* packed_ptr);
[all …]
|
| D | pack_x86.h | 61 int remaining_src_cols, int src_rows,
108 int remaining_src_cols, int src_rows,
155 int src_rows, float* packed_ptr);
189 int src_rows, float* packed_ptr);
226 int remaining_src_cols, int src_rows,
276 int src_rows, float* packed_ptr);
311 int src_cols, int block_row, int src_rows,
346 int src_cols, int block_row, int src_rows,
381 int src_cols, int block_row, int src_rows,
469 int src_rows, float* packed_ptr,
[all …]
|
| /external/tensorflow/tensorflow/core/kernels/linalg/ |
| D | tridiagonal_solve_op_gpu.cu.cc | 188 const Scalar* src, Scalar* dst, const int src_rows, in TransposeWithGeam() argument
192 cublas_solver->Geam(CUBLAS_OP_T, CUBLAS_OP_N, src_rows, in TransposeWithGeam()
195 src_rows, dst, src_rows)); in TransposeWithGeam()
|
| /external/libaom/libaom/av1/common/ |
| D | restoration.c | 1410 const uint8_t *src_rows = src_buf + row * src_stride; in save_deblock_boundary_lines() local
1435 cm, CONVERT_TO_BYTEPTR(src_rows), frame->strides[is_uv], in save_deblock_boundary_lines()
1439 av1_upscale_normative_rows(cm, src_rows, frame->strides[is_uv], bdry_rows, in save_deblock_boundary_lines()
1446 memcpy(bdry_rows + i * bdry_stride, src_rows + i * src_stride, in save_deblock_boundary_lines()
1465 const uint8_t *src_rows = src_buf + row * src_stride; in save_cdef_boundary_lines() local
1487 memcpy(bdry_rows + i * bdry_stride, src_rows, line_bytes); in save_cdef_boundary_lines()
|