| /third_party/python/Modules/_blake2/impl/ |
| D | blake2s-round.h | 44 #define G1(row1,row2,row3,row4,buf) \ argument
46 row4 = _mm_xor_si128( row4, row1 ); \
47 row4 = _mm_roti_epi32(row4, -16); \
48 row3 = _mm_add_epi32( row3, row4 ); \
52 #define G2(row1,row2,row3,row4,buf) \ argument
54 row4 = _mm_xor_si128( row4, row1 ); \
55 row4 = _mm_roti_epi32(row4, -8); \
56 row3 = _mm_add_epi32( row3, row4 ); \
60 #define DIAGONALIZE(row1,row2,row3,row4) \ argument
61 row4 = _mm_shuffle_epi32( row4, _MM_SHUFFLE(2,1,0,3) ); \
[all …]
|
| D | blake2s.c | 268 __m128i row1, row2, row3, row4; in blake2s_compress() local
307 …row4 = _mm_xor_si128( _mm_setr_epi32( 0x510E527F, 0x9B05688C, 0x1F83D9AB, 0x5BE0CD19 ), LOADU( &S-… in blake2s_compress()
319 STOREU( &S->h[4], _mm_xor_si128( ff1, _mm_xor_si128( row2, row4 ) ) ); in blake2s_compress()
|
| /third_party/libjpeg-turbo/simd/powerpc/ |
| D | jquanti-altivec.c | 119 __vector short row0, row1, row2, row3, row4, row5, row6, row7, in jsimd_quantize_altivec() local
140 row4 = vec_ld(64, workspace); in jsimd_quantize_altivec()
150 row4s = vec_sra(row4, pw_word_bit_m1); in jsimd_quantize_altivec()
158 row4 = vec_xor(row4, row4s); in jsimd_quantize_altivec()
166 row4 = vec_sub(row4, row4s); in jsimd_quantize_altivec()
184 row4 = vec_add(row4, corr4); in jsimd_quantize_altivec()
202 MULTIPLY(row4, recip4, row4); in jsimd_quantize_altivec()
220 MULTIPLY(row4, scale4, row4); in jsimd_quantize_altivec()
229 row4 = vec_xor(row4, row4s); in jsimd_quantize_altivec()
237 row4 = vec_sub(row4, row4s); in jsimd_quantize_altivec()
[all …]
|
| D | jfdctfst-altivec.c | 92 __vector short row0, row1, row2, row3, row4, row5, row6, row7, in jsimd_fdct_ifast_altivec() local
113 row4 = vec_ld(64, data); in jsimd_fdct_ifast_altivec()
141 tmp3 = vec_add(row3, row4); in jsimd_fdct_ifast_altivec()
142 tmp4 = vec_sub(row3, row4); in jsimd_fdct_ifast_altivec()
|
| D | jfdctint-altivec.c | 184 __vector short row0, row1, row2, row3, row4, row5, row6, row7, in jsimd_fdct_islow_altivec() local
217 row4 = vec_ld(64, data); in jsimd_fdct_islow_altivec()
245 tmp3 = vec_add(row3, row4); in jsimd_fdct_islow_altivec()
246 tmp4 = vec_sub(row3, row4); in jsimd_fdct_islow_altivec()
|
| /third_party/flutter/skia/third_party/externals/libjpeg-turbo/simd/powerpc/ |
| D | jquanti-altivec.c | 119 __vector short row0, row1, row2, row3, row4, row5, row6, row7, in jsimd_quantize_altivec() local
140 row4 = vec_ld(64, workspace); in jsimd_quantize_altivec()
150 row4s = vec_sra(row4, pw_word_bit_m1); in jsimd_quantize_altivec()
158 row4 = vec_xor(row4, row4s); in jsimd_quantize_altivec()
166 row4 = vec_sub(row4, row4s); in jsimd_quantize_altivec()
184 row4 = vec_add(row4, corr4); in jsimd_quantize_altivec()
202 MULTIPLY(row4, recip4, row4); in jsimd_quantize_altivec()
220 MULTIPLY(row4, scale4, row4); in jsimd_quantize_altivec()
229 row4 = vec_xor(row4, row4s); in jsimd_quantize_altivec()
237 row4 = vec_sub(row4, row4s); in jsimd_quantize_altivec()
[all …]
|
| D | jfdctfst-altivec.c | 92 __vector short row0, row1, row2, row3, row4, row5, row6, row7, in jsimd_fdct_ifast_altivec() local
113 row4 = vec_ld(64, data); in jsimd_fdct_ifast_altivec()
141 tmp3 = vec_add(row3, row4); in jsimd_fdct_ifast_altivec()
142 tmp4 = vec_sub(row3, row4); in jsimd_fdct_ifast_altivec()
|
| D | jfdctint-altivec.c | 184 __vector short row0, row1, row2, row3, row4, row5, row6, row7, in jsimd_fdct_islow_altivec() local
217 row4 = vec_ld(64, data); in jsimd_fdct_islow_altivec()
245 tmp3 = vec_add(row3, row4); in jsimd_fdct_islow_altivec()
246 tmp4 = vec_sub(row3, row4); in jsimd_fdct_islow_altivec()
|
| /third_party/skia/third_party/externals/libjpeg-turbo/simd/arm/aarch32/ |
| D | jchuff-neon.c | 138 int16x8_t row4 = vld1q_dup_s16(block + 35); in jsimd_huff_encode_one_block_neon() local
139 row4 = vld1q_lane_s16(block + 42, row4, 1); in jsimd_huff_encode_one_block_neon()
140 row4 = vld1q_lane_s16(block + 49, row4, 2); in jsimd_huff_encode_one_block_neon()
141 row4 = vld1q_lane_s16(block + 56, row4, 3); in jsimd_huff_encode_one_block_neon()
142 row4 = vld1q_lane_s16(block + 57, row4, 4); in jsimd_huff_encode_one_block_neon()
143 row4 = vld1q_lane_s16(block + 50, row4, 5); in jsimd_huff_encode_one_block_neon()
144 row4 = vld1q_lane_s16(block + 43, row4, 6); in jsimd_huff_encode_one_block_neon()
145 row4 = vld1q_lane_s16(block + 36, row4, 7); in jsimd_huff_encode_one_block_neon()
174 int16x8_t abs_row4 = vabsq_s16(row4); in jsimd_huff_encode_one_block_neon()
200 vshlq_u16(vreinterpretq_u16_s16(vshrq_n_s16(row4, 15)), in jsimd_huff_encode_one_block_neon()
|
| /third_party/mindspore/mindspore/lite/micro/coder/wrapper/base/ |
| D | optimize_handler_wrapper.c | 19 extern void MatMulOptR4Int8Neon64(const int8_t *a, const int8_t *b, int *dst, int row4, int col4, i…
31 void MatMulR4Int8_optimize_handler(const int8_t *a, const int8_t *b, int *dst, int row4, int col4, … in MatMulR4Int8_optimize_handler() argument
33 return MatMulOptR4Int8Neon64(a, b, dst, row4, col4, deep16, input_sum, bias); in MatMulR4Int8_optimize_handler()
|
| /third_party/skia/third_party/externals/libjpeg-turbo/simd/arm/ |
| D | jfdctfst-neon.c | 156 int16x8_t row4 = vreinterpretq_s16_s32(rows_04.val[1]); in jsimd_fdct_ifast_neon() local
169 tmp3 = vaddq_s16(row3, row4); in jsimd_fdct_ifast_neon()
170 tmp4 = vsubq_s16(row3, row4); in jsimd_fdct_ifast_neon()
179 row4 = vsubq_s16(tmp10, tmp11); in jsimd_fdct_ifast_neon()
210 vst1q_s16(data + 4 * DCTSIZE, row4); in jsimd_fdct_ifast_neon()
|
| D | jidctfst-neon.c | 73 int16x8_t row4 = vld1q_s16(coef_block + 4 * DCTSIZE); in jsimd_idct_ifast_neon() local
86 bitmap = vorrq_s16(bitmap, row4); in jsimd_idct_ifast_neon()
105 row4 = dcval; in jsimd_idct_ifast_neon()
129 int16x4_t tmp2 = vmul_s16(vget_high_s16(row4), quant_row4); in jsimd_idct_ifast_neon()
182 row4 = vcombine_s16(dcval, vadd_s16(tmp3, tmp4)); in jsimd_idct_ifast_neon()
204 int16x4_t tmp2 = vmul_s16(vget_low_s16(row4), quant_row4); in jsimd_idct_ifast_neon()
257 row4 = vcombine_s16(vadd_s16(tmp3, tmp4), dcval); in jsimd_idct_ifast_neon()
274 int16x8_t tmp2 = vmulq_s16(row4, quant_row4); in jsimd_idct_ifast_neon()
327 row4 = vaddq_s16(tmp3, tmp4); in jsimd_idct_ifast_neon()
334 int16x8x2_t rows_45 = vtrnq_s16(row4, row5); in jsimd_idct_ifast_neon()
|
| D | jidctint-neon.c | 101 int16x4_t row4,
205 int16x4_t row4 = vld1_s16(coef_block + 4 * DCTSIZE); in jsimd_idct_islow_neon() local
223 bitmap = vorr_s16(bitmap, row4); in jsimd_idct_islow_neon()
244 jsimd_idct_islow_pass1_regular(row0, row1, row2, row3, row4, row5, in jsimd_idct_islow_neon()
258 row4 = vld1_s16(coef_block + 4 * DCTSIZE + 4); in jsimd_idct_islow_neon()
276 bitmap = vorr_s16(bitmap, row4); in jsimd_idct_islow_neon()
304 jsimd_idct_islow_pass1_regular(row0, row1, row2, row3, row4, row5, in jsimd_idct_islow_neon()
342 int16x4_t row4, in jsimd_idct_islow_pass1_regular() argument
377 z3_s16 = vmul_s16(row4, quant_row4); in jsimd_idct_islow_pass1_regular()
|
| D | jfdctint-neon.c | 255 int16x8_t row4 = vreinterpretq_s16_s32(rows_04.val[1]); in jsimd_fdct_islow_neon() local
268 tmp3 = vaddq_s16(row3, row4); in jsimd_fdct_islow_neon()
269 tmp4 = vsubq_s16(row3, row4); in jsimd_fdct_islow_neon()
278 row4 = vrshrq_n_s16(vsubq_s16(tmp10, tmp11), PASS1_BITS); in jsimd_fdct_islow_neon()
372 vst1q_s16(data + 4 * DCTSIZE, row4); in jsimd_fdct_islow_neon()
|
| /third_party/ffmpeg/libavcodec/mips/ |
| D | vp8_lpf_msa.c | 380 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in ff_vp8_h_loop_filter16_msa() local
388 LD_UB8(temp_src, pitch, row0, row1, row2, row3, row4, row5, row6, row7); in ff_vp8_h_loop_filter16_msa()
392 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in ff_vp8_h_loop_filter16_msa()
445 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in ff_vp8_h_loop_filter8uv_msa() local
453 LD_UB8(src_u - 4, pitch, row0, row1, row2, row3, row4, row5, row6, row7); in ff_vp8_h_loop_filter8uv_msa()
456 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in ff_vp8_h_loop_filter8uv_msa()
525 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in ff_vp8_h_loop_filter_simple_msa() local
531 LD_UB8(temp_src, pitch, row0, row1, row2, row3, row4, row5, row6, row7); in ff_vp8_h_loop_filter_simple_msa()
535 TRANSPOSE16x4_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in ff_vp8_h_loop_filter_simple_msa()
595 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in ff_vp8_h_loop_filter8uv_inner_msa() local
[all …]
|
| /third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/int8/ |
| D | opt_op_handler.cc | 27 void MatMulR4Int8_optimize_handler(const int8_t *a, const int8_t *b, int *dst, int row4, int col4, … in MatMulR4Int8_optimize_handler() argument
29 return MatMulOptR4Int8Neon64(a, b, dst, row4, col4, deep16, input_sum, bias); in MatMulR4Int8_optimize_handler()
|
| D | opt_op_handler.h | 25 void MatMulOptR4Int8Neon64(const int8_t *a, const int8_t *b, int *dst, int row4, int col4, int deep…
39 void MatMulR4Int8_optimize_handler(const int8_t *a, const int8_t *b, int *dst, int row4, int col4, …
|
| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/int8/ |
| D | pack_int8.h | 41 void PackInputSum16x4PerLayer(const int8_t *src, int32_t *dst, int32_t filter_zp, size_t row4, size…
48 void PreSum4x16Int8Pert(const int8_t *src, int32_t *sum, size_t row4, size_t col16, int32_t filter_…
|
| D | matmul_int8.h | 69 void MatmulInt8Neon64(const int8_t *a, const int8_t *b, int8_t *dst, int row4, int col4, int deep16…
73 void MatMulR4Int8Neon64(const int8_t *a, const int8_t *b, int32_t *dst, int row4, int col4, int dee…
|
| /third_party/flutter/skia/gm/ |
| D | hardstop_gradients.cpp | 118 SkScalar row4[] = {0.00f, 0.25f, 0.50f, 0.50f, 1.00f}; in onDraw() local
128 row4, in onDraw()
|
| /third_party/skia/gm/ |
| D | hardstop_gradients.cpp | 118 SkScalar row4[] = {0.00f, 0.25f, 0.50f, 0.50f, 1.00f}; in onDraw() local
128 row4, in onDraw()
|
| /third_party/skia/third_party/externals/d3d12allocator/src/ |
| D | Common.h | 187 const vec4& row4) : in mat4()
191 _41(row4.x), _42(row4.y), _43(row4.z), _44(row4.w) in mat4()
|
| /third_party/flutter/skia/third_party/externals/libwebp/src/dsp/ |
| D | dec_msa.c | 364 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in HFilter16() local
371 LD_UB8(ptmp, stride, row0, row1, row2, row3, row4, row5, row6, row7); in HFilter16()
374 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in HFilter16()
446 v16u8 row0, row1, row2, row3, row4, row5, row6, row7; in HFilterVertEdge16i() local
453 LD_UB8(src - 4, stride, row0, row1, row2, row3, row4, row5, row6, row7); in HFilterVertEdge16i()
456 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in HFilterVertEdge16i()
530 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in HFilter8() local
537 LD_UB8(ptmp_src_u, stride, row0, row1, row2, row3, row4, row5, row6, row7); in HFilter8()
540 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in HFilter8()
595 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in HFilter8i() local
[all …]
|
| /third_party/skia/third_party/externals/libwebp/src/dsp/ |
| D | dec_msa.c | 364 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in HFilter16() local
371 LD_UB8(ptmp, stride, row0, row1, row2, row3, row4, row5, row6, row7); in HFilter16()
374 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in HFilter16()
446 v16u8 row0, row1, row2, row3, row4, row5, row6, row7; in HFilterVertEdge16i() local
453 LD_UB8(src - 4, stride, row0, row1, row2, row3, row4, row5, row6, row7); in HFilterVertEdge16i()
456 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in HFilterVertEdge16i()
530 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in HFilter8() local
537 LD_UB8(ptmp_src_u, stride, row0, row1, row2, row3, row4, row5, row6, row7); in HFilter8()
540 TRANSPOSE16x8_UB_UB(row0, row1, row2, row3, row4, row5, row6, row7, in HFilter8()
595 v16u8 row0, row1, row2, row3, row4, row5, row6, row7, row8; in HFilter8i() local
[all …]
|
| /third_party/ffmpeg/libavcodec/ |
| D | ivi_dsp.c | 669 int i, row2, row4, row8; in ff_ivi_col_slant8() local
673 row4 = pitch << 2; in ff_ivi_col_slant8()
680 out[0], out[pitch], out[row2], out[row2 + pitch], out[row4], in ff_ivi_col_slant8()
681 out[row4 + pitch], out[row4 + row2], out[row8 - pitch], in ff_ivi_col_slant8()
684 out[0] = out[pitch] = out[row2] = out[row2 + pitch] = out[row4] = in ff_ivi_col_slant8()
685 out[row4 + pitch] = out[row4 + row2] = out[row8 - pitch] = 0; in ff_ivi_col_slant8()
|