/external/tensorflow/third_party/eigen3/unsupported/Eigen/CXX11/src/FixedPoint/ |
D | PacketMathAVX2.h | 439 _mm256_min_epi32(tmp, _mm256_shuffle_epi32(tmp, _MM_SHUFFLE(1, 0, 3, 2))); 441 _mm256_min_epi32(tmp, _mm256_shuffle_epi32(tmp, 1))); 447 _mm256_max_epi32(tmp, _mm256_shuffle_epi32(tmp, _MM_SHUFFLE(1, 0, 3, 2))); 449 _mm256_max_epi32(tmp, _mm256_shuffle_epi32(tmp, 1))); 456 _mm256_min_epi16(tmp, _mm256_shuffle_epi32(tmp, _MM_SHUFFLE(1, 0, 3, 2))); 457 tmp = _mm256_min_epi16(tmp, _mm256_shuffle_epi32(tmp, 1)); 464 _mm256_max_epi16(tmp, _mm256_shuffle_epi32(tmp, _MM_SHUFFLE(1, 0, 3, 2))); 465 tmp = _mm256_max_epi16(tmp, _mm256_shuffle_epi32(tmp, 1)); 473 _mm256_min_epu8(tmp, _mm256_shuffle_epi32(tmp, _MM_SHUFFLE(1, 0, 3, 2))); 474 tmp = _mm256_min_epu8(tmp, _mm256_shuffle_epi32(tmp, 1)); [all …]
|
/external/libaom/libaom/av1/encoder/x86/ |
D | encodetxb_avx2.c | 47 const __m256i res_ = _mm256_shuffle_epi32(abs01_8, 0xd8); in av1_txb_init_levels_avx2() 66 const __m256i res = _mm256_shuffle_epi32(res_, 0xd8); in av1_txb_init_levels_avx2() 93 const __m256i res = _mm256_shuffle_epi32(res_, 0xd8); in av1_txb_init_levels_avx2() 114 const __m256i res = _mm256_shuffle_epi32(res_, 0xd8); in av1_txb_init_levels_avx2()
|
D | av1_highbd_quantize_avx2.c | 127 eob_s = _mm256_shuffle_epi32(eob, 0xe); in av1_highbd_quantize_fp_avx2()
|
/external/ruy/ruy/ |
D | kernel_avx2_fma.cc | 313 tmp0 = _mm256_shuffle_epi32(rhs0_3, 0); 314 tmp1 = _mm256_shuffle_epi32(rhs0_3, 0x55); 316 tmp2 = _mm256_shuffle_epi32(rhs0_3, 0xaa); 317 tmp3 = _mm256_shuffle_epi32(rhs0_3, 0xff); 320 tmp0 = _mm256_shuffle_epi32(rhs4_7, 0); 321 tmp1 = _mm256_shuffle_epi32(rhs4_7, 0x55); 323 tmp2 = _mm256_shuffle_epi32(rhs4_7, 0xaa); 324 tmp3 = _mm256_shuffle_epi32(rhs4_7, 0xff); 327 tmp0 = _mm256_shuffle_epi32(rhs8_11, 0); 328 tmp1 = _mm256_shuffle_epi32(rhs8_11, 0x55); [all …]
|
/external/libaom/libaom/av1/common/x86/ |
D | wiener_convolve_avx2.c | 107 coeffs_v[0] = _mm256_shuffle_epi32(filter_coeffs_y, 0x00); in av1_wiener_convolve_add_src_avx2() 109 coeffs_v[1] = _mm256_shuffle_epi32(filter_coeffs_y, 0x55); in av1_wiener_convolve_add_src_avx2() 111 coeffs_v[2] = _mm256_shuffle_epi32(filter_coeffs_y, 0xaa); in av1_wiener_convolve_add_src_avx2() 113 coeffs_v[3] = _mm256_shuffle_epi32(filter_coeffs_y, 0xff); in av1_wiener_convolve_add_src_avx2()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 3x8c8-minmax-avx2.c | 156 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2() 157 const __m256i vacc1x11335577 = _mm256_shuffle_epi32(vacc1x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2() 158 const __m256i vacc2x11335577 = _mm256_shuffle_epi32(vacc2x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
|
D | 2x8c8-minmax-avx2.c | 133 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2() 134 const __m256i vacc1x11335577 = _mm256_shuffle_epi32(vacc1x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2()
|
D | 1x8c8-minmax-avx2.c | 110 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_igemm_minmax_ukernel_1x8c8__avx2()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x8c8-minmax-avx2.c | 139 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2() 140 const __m256i vacc1x11335577 = _mm256_shuffle_epi32(vacc1x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2() 141 const __m256i vacc2x11335577 = _mm256_shuffle_epi32(vacc2x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
|
D | 3x8c8-xw-minmax-avx2.c | 135 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2() 136 const __m256i vacc1x11335577 = _mm256_shuffle_epi32(vacc1x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2() 137 const __m256i vacc2x11335577 = _mm256_shuffle_epi32(vacc2x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
|
D | 2x8c8-minmax-avx2.c | 118 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2() 119 const __m256i vacc1x11335577 = _mm256_shuffle_epi32(vacc1x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2()
|
D | 2x8c8-xw-minmax-avx2.c | 114 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2() 115 const __m256i vacc1x11335577 = _mm256_shuffle_epi32(vacc1x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2()
|
D | 1x8c8-minmax-avx2.c | 97 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_minmax_ukernel_1x8c8__avx2()
|
D | 1x8c8-xw-minmax-avx2.c | 93 const __m256i vacc0x11335577 = _mm256_shuffle_epi32(vacc0x01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gemm_xw_minmax_ukernel_1x8c8__avx2()
|
/external/libaom/libaom/aom_dsp/x86/ |
D | convolve_avx2.h | 308 coeffs[0] = _mm256_shuffle_epi32(coeff, 0x00); in prepare_coeffs() 310 coeffs[1] = _mm256_shuffle_epi32(coeff, 0x55); in prepare_coeffs() 312 coeffs[2] = _mm256_shuffle_epi32(coeff, 0xaa); in prepare_coeffs() 314 coeffs[3] = _mm256_shuffle_epi32(coeff, 0xff); in prepare_coeffs()
|
D | masked_sad_intrin_avx2.c | 62 res = _mm256_shuffle_epi32(res, 0xd8); in masked_sad32xh_avx2() 115 res = _mm256_shuffle_epi32(res, 0xd8); in masked_sad16xh_avx2()
|
D | highbd_quantize_intrin_avx2.c | 150 eob_s = _mm256_shuffle_epi32(eob, 0xe); in aom_highbd_quantize_b_avx2()
|
/external/XNNPACK/src/qs8-dwconv/gen/ |
D | up32x9-minmax-avx2-mul16.c | 229 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16() 230 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16() 231 const __m256i vaccHJLN = _mm256_shuffle_epi32(vaccGHIJKLMN, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16() 232 const __m256i vaccPRTV = _mm256_shuffle_epi32(vaccOPQRSTUV, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16() 388 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16() 389 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
|
D | up16x9-minmax-avx2-mul16.c | 173 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16() 174 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16() 296 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16() 297 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16()
|
D | up32x9-minmax-avx2-mul32.c | 229 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32() 230 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32() 231 const __m256i vaccHJLN = _mm256_shuffle_epi32(vaccGHIJKLMN, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32() 232 const __m256i vaccPRTV = _mm256_shuffle_epi32(vaccOPQRSTUV, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32() 360 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32()
|
D | up24x9-minmax-avx2-mul32.c | 201 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32() 202 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32() 203 const __m256i vaccHJLN = _mm256_shuffle_epi32(vaccGHIJKLMN, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32() 322 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32()
|
D | up16x9-minmax-avx2-mul32.c | 173 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul32() 174 const __m256i vacc9BDF = _mm256_shuffle_epi32(vacc89ABCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul32() 280 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul32()
|
D | up8x9-minmax-avx2-mul32.c | 145 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up8x9__avx2_mul32() 230 const __m256i vacc1357 = _mm256_shuffle_epi32(vacc01234567, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up8x9__avx2_mul32()
|
/external/libvpx/libvpx/vpx_dsp/x86/ |
D | highbd_convolve_avx2.c | 966 kernel_reg_23 = _mm256_shuffle_epi32(kernel_reg, 0x55); in vpx_highbd_filter_block1d4_h4_avx2() 967 kernel_reg_45 = _mm256_shuffle_epi32(kernel_reg, 0xaa); in vpx_highbd_filter_block1d4_h4_avx2() 1053 kernel_reg_23 = _mm256_shuffle_epi32(kernel_reg, 0x55); in vpx_highbd_filter_block1d8_h4_avx2() 1054 kernel_reg_45 = _mm256_shuffle_epi32(kernel_reg, 0xaa); in vpx_highbd_filter_block1d8_h4_avx2() 1282 kernel_reg_23 = _mm256_shuffle_epi32(kernel_reg, 0x55); in vpx_highbd_filter_block1d4_v4_avx2() 1283 kernel_reg_45 = _mm256_shuffle_epi32(kernel_reg, 0xaa); in vpx_highbd_filter_block1d4_v4_avx2() 1367 kernel_reg_23 = _mm256_shuffle_epi32(kernel_reg, 0x55); in vpx_highbd_filter_block1d8_v4_avx2() 1368 kernel_reg_45 = _mm256_shuffle_epi32(kernel_reg, 0xaa); in vpx_highbd_filter_block1d8_v4_avx2()
|
/external/XNNPACK/src/qs8-dwconv/ |
D | unipass-avx2-mul16.c.in | 70 …const __m256i vacc${ABC[C+1:C+8:2]} = _mm256_shuffle_epi32(vacc${ABC[C:C+8]}, _MM_SHUFFLE(3, 3, 1,… 143 … const __m256i vacc${ABC[1:8:2]} = _mm256_shuffle_epi32(vacc${ABC[0:8]}, _MM_SHUFFLE(3, 3, 1, 1)); 144 …const __m256i vacc${ABC[9:16:2]} = _mm256_shuffle_epi32(vacc${ABC[8:16]}, _MM_SHUFFLE(3, 3, 1, 1));
|