/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp32/ |
D | matmul_fp32.c | 142 void RowMajor2Col12Major_arm64(const float *src_c, float *dst_c, size_t col) { in RowMajor2Col12Major_arm64() argument 198 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) in RowMajor2Col12Major_arm64() 206 void RowMajor2Col12Major_arm32(const float *src_c, float *dst_c, size_t col) { in RowMajor2Col12Major_arm32() argument 259 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) in RowMajor2Col12Major_arm32() 272 float *dst_c = dst_r + ci * C12NUM; in RowMajor2Col12Major() local 274 RowMajor2Col12Major_arm64(src_c, dst_c, col); in RowMajor2Col12Major() 276 RowMajor2Col12Major_arm32(src_c, dst_c, col); in RowMajor2Col12Major() 321 _mm_storeu_ps(dst_c, dst0); in RowMajor2Col12Major() 322 _mm_storeu_ps(dst_c + 4, dst1); in RowMajor2Col12Major() 323 _mm_storeu_ps(dst_c + 8, dst2); in RowMajor2Col12Major() [all …]
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp32_grad/ |
D | gemm.c | 105 float *dst_c = dst_r + ci * C12NUM; local 164 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) 221 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) 226 dst_c[tc * C12NUM + tr] = src_c[tr * lead + tc]; 233 float *dst_c = dst_r + ci * C12NUM; local 235 dst_c[i] = src_c[i * lead]; 276 float *dst_c = dst_r + ci * C8NUM; local 353 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) 397 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) 438 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ stride ] "r"(stride) [all …]
|
/third_party/openh264/test/processing/ |
D | ProcessUT_DownSample.cpp | 194 ENFORCE_STACK_ALIGN_1D (uint8_t, dst_c, 50000, 16); \ 211 dst_c[j] = dst_a[j] = rand() % 256; \ 214 ref_func (dst_c, dst_stride_c, src_c, src_stride_c, src_width_c, src_height_c); \ 218 ASSERT_EQ (dst_c[m + j * dst_stride_c], dst_a[m + j * dst_stride_a]); \ 236 ENFORCE_STACK_ALIGN_1D (uint8_t, dst_c, 50000, 16); \ 253 dst_c[j] = dst_a[j] = rand() % 256; \ 256 …DyadicBilinearOneThirdDownsampler_c (dst_c, dst_stride_c, src_c, src_stride_c, src_width_c, src_he… 260 ASSERT_EQ (dst_c[m + j * dst_stride_c], dst_a[m + j * dst_stride_a]); \ 273 ENFORCE_STACK_ALIGN_1D (uint8_t, dst_c, 50000, 16); \ 290 dst_c[j] = dst_a[j] = rand() % 256; \ [all …]
|
/third_party/mindspore/mindspore/lite/src/runtime/kernel/opencl/cl/ |
D | conv2d_transpose.cl | 31 int dst_c = get_global_id(2); // n * c4 32 int n = dst_c / dst_size.z; 33 dst_c = dst_c % dst_size.z; 34 if (dst_h >= dst_size.x || dst_w >= dst_size.y || dst_c >= dst_size.z || n >= dst_size.w) return; 35 int weight_base = dst_c * src_size.z * kernel_size.x * kernel_size.y; 95 FLT4 bias_val = READ_IMAGE(biases, smp_zero, (int2)(dst_c, 0)); 113 WRITE_IMAGE(dst_data, (int2)(dst_w * dst_size.z + dst_c, n * dst_size.x + dst_h), r0); 115 … WRITE_IMAGE(dst_data, (int2)(dst_w * dst_size.z + dst_c, n * dst_size.x + dst_h + stride.x), r1); 118 …WRITE_IMAGE(dst_data, (int2)((dst_w + stride.y) * dst_size.z + dst_c, n * dst_size.x + dst_h), r2); 121 …WRITE_IMAGE(dst_data, (int2)((dst_w + stride.y) * dst_size.z + dst_c, n * dst_size.x + dst_h + str…
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp16_grad/ |
D | gemm_fp16.c | 128 : [ dst_c ] "r"(dst_ptr), [ src_c ] "r"(src_ptr), [ stride ] "r"(stride) in Row2Col16Block16() 188 float16_t *dst_c = dst_r + ci * C16NUM; in RowMajor2Col16MajorStrideFp16() local 190 Row2Col16Block16(src_c, dst_c, stride); in RowMajor2Col16MajorStrideFp16() 194 dst_c[tc * C16NUM + tr] = src_c[tr * stride + tc]; in RowMajor2Col16MajorStrideFp16() 201 float16_t *dst_c = dst_r + ci * C16NUM; in RowMajor2Col16MajorStrideFp16() local 203 dst_c[i] = src_c[i * stride]; in RowMajor2Col16MajorStrideFp16() 248 float16_t *dst_c = dst_r + ci * C12NUM; in RowMajor2Col12MajorStrideFp16() local 250 Transpose12x8A32Fp16(src_c, dst_c, stride * sizeof(float16_t), 24); in RowMajor2Col12MajorStrideFp16() 254 dst_c[tc * C12NUM + tr] = src_c[tr * stride + tc]; in RowMajor2Col12MajorStrideFp16() 261 float16_t *dst_c = dst_r + ci * C12NUM; in RowMajor2Col12MajorStrideFp16() local [all …]
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp16/ |
D | matmul_fp16.c | 600 : [ dst_c ] "r"(dst_ptr), [ src_c ] "r"(src_ptr), [ stride ] "r"(stride) in Row2Col16Block16() 619 float16_t *dst_c = dst_r + ci * C16NUM; in RowMajor2Col16MajorFp16Opt() local 621 Row2Col16Block16(src_c, dst_c, col); in RowMajor2Col16MajorFp16Opt() 625 dst_c[tc * C16NUM + tr] = src_c[tr * col + tc]; in RowMajor2Col16MajorFp16Opt() 632 float16_t *dst_c = dst_r + ci * C16NUM; in RowMajor2Col16MajorFp16Opt() local 634 dst_c[i] = src_c[i * col]; in RowMajor2Col16MajorFp16Opt() 668 float16_t *dst_c = dst_r + ci * C16NUM; in RowMajor2ColNMajorFp16() local 669 Row2Col16Block16(src_c, dst_c, col); in RowMajor2ColNMajorFp16() 673 float16_t *dst_c = dst_r + ci * C16NUM; in RowMajor2ColNMajorFp16() local 675 dst_c[i] = src_c[i * col]; in RowMajor2ColNMajorFp16() [all …]
|
D | pack_fp16.c | 584 inline void Transpose12x8A32Fp16(const float16_t *src_c, float16_t *dst_c, size_t src_stride, size_… in Transpose12x8A32Fp16() argument 655 …: [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ src_stride ] "r"(src_stride), [ dst_stride ] "r"(d… in Transpose12x8A32Fp16()
|
/third_party/mesa3d/src/compiler/isaspec/ |
D | decode.py | 296 dst_c = sys.argv[3] variable 305 with open(dst_c, 'w') as f:
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/int8/ |
D | matmul_int8.c | 109 int8_t *dst_c = dst_r + ci * C4NUM; in RowMajor2Row16x4MajorInt8() local 126 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ col_offset ] "r"(col_offset) in RowMajor2Row16x4MajorInt8() 146 : [ dst_c ] "r"(dst_c), [ src_c ] "r"(src_c), [ col_offset ] "r"(col_offset) in RowMajor2Row16x4MajorInt8() 149 MatrixPack4x16UnitInt8(src_c, dst_c, C4NUM, C16NUM, col_offset); in RowMajor2Row16x4MajorInt8()
|
D | pack_int8.c | 975 int16_t *dst_c = packed_weight_ + c8_block_num * plane * C8NUM; in PackDepthwiseInt8Weight() local 978 int16_t *dst_kernel = dst_c + C8NUM * k + c8_block_rem; in PackDepthwiseInt8Weight() 994 int16_t *dst_c = packed_weight_ + c4_block_num * plane * C4NUM; in PackDeconvDepthwiseInt8Weight() local 997 int16_t *dst_kernel = dst_c + C4NUM * k + c4_block_rem; in PackDeconvDepthwiseInt8Weight()
|
/third_party/mindspore/mindspore/ccsrc/minddata/dataset/kernels/image/lite_cv/ |
D | image_process.cc | 639 int dst_c = src.channel_; in CropInternal() local 641 dst.Init(dst_w, dst_h, dst_c, src.data_type_); in CropInternal() 650 const T *src_index_p = src_start_p + (y + i_h) * src.width_ * dst_c + x * dst_c; in CropInternal() 651 T *dst_index_p = dst_start_p + i_h * dst_w * dst_c; in CropInternal() 653 (void)memcpy(dst_index_p, src_index_p, dst_w * dst_c * sizeof(T)); in CropInternal()
|