| /external/libaom/libaom/av1/encoder/x86/ |
| D | highbd_block_error_intrin_avx2.c | 35 __m256i diff1h = _mm256_srli_epi64(diff1, 32); in av1_highbd_block_error_avx2()
36 __m256i diff2h = _mm256_srli_epi64(diff2, 32); in av1_highbd_block_error_avx2()
43 __m256i mm256_coeffh = _mm256_srli_epi64(mm256_coeff, 32); in av1_highbd_block_error_avx2()
44 __m256i mm256_coeffh2 = _mm256_srli_epi64(mm256_coeff2, 32); in av1_highbd_block_error_avx2()
|
| D | av1_highbd_quantize_avx2.c | 56 __m256i q_hi = _mm256_srli_epi64(q, 32); in quantize()
57 const __m256i qp_hi = _mm256_srli_epi64(qp[1], 32); in quantize()
59 q_lo = _mm256_srli_epi64(q_lo, 16 - log_scale); in quantize()
60 q_hi = _mm256_srli_epi64(q_hi, 16 - log_scale); in quantize()
|
| D | pickrst_avx2.c | 661 const __m256i h00_odd = _mm256_mul_epi32(_mm256_srli_epi64(f1, 32), in calc_proj_params_r0_r1_avx2()
662 _mm256_srli_epi64(f1, 32)); in calc_proj_params_r0_r1_avx2()
667 const __m256i h01_odd = _mm256_mul_epi32(_mm256_srli_epi64(f1, 32), in calc_proj_params_r0_r1_avx2()
668 _mm256_srli_epi64(f2, 32)); in calc_proj_params_r0_r1_avx2()
673 const __m256i h11_odd = _mm256_mul_epi32(_mm256_srli_epi64(f2, 32), in calc_proj_params_r0_r1_avx2()
674 _mm256_srli_epi64(f2, 32)); in calc_proj_params_r0_r1_avx2()
680 _mm256_mul_epi32(_mm256_srli_epi64(f1, 32), _mm256_srli_epi64(s, 32)); in calc_proj_params_r0_r1_avx2()
686 _mm256_mul_epi32(_mm256_srli_epi64(f2, 32), _mm256_srli_epi64(s, 32)); in calc_proj_params_r0_r1_avx2()
754 const __m256i h00_odd = _mm256_mul_epi32(_mm256_srli_epi64(f1, 32), in calc_proj_params_r0_avx2()
755 _mm256_srli_epi64(f1, 32)); in calc_proj_params_r0_avx2()
[all …]
|
| D | wedge_utils_avx2.c | 64 _mm256_and_si256(v_sq0_d, v_zext_q), _mm256_srli_epi64(v_sq0_d, 32)); in av1_wedge_sse_from_residuals_avx2()
|
| /external/libaom/libaom/aom_dsp/x86/ |
| D | highbd_quantize_intrin_avx2.c | 53 __m256i prod_hi = _mm256_srli_epi64(*x, 32); in mm256_mul_shift_epi32()
54 const __m256i mult_hi = _mm256_srli_epi64(*y, 32); in mm256_mul_shift_epi32()
57 prod_lo = _mm256_srli_epi64(prod_lo, 16); in mm256_mul_shift_epi32()
60 prod_hi = _mm256_srli_epi64(prod_hi, 16); in mm256_mul_shift_epi32()
|
| D | highbd_adaptive_quantize_avx2.c | 64 __m256i prod_hi = _mm256_srli_epi64(*x, 32); in highbd_mul_shift_avx2()
65 const __m256i mult_hi = _mm256_srli_epi64(*y, 32); in highbd_mul_shift_avx2()
68 prod_lo = _mm256_srli_epi64(prod_lo, shift); in highbd_mul_shift_avx2()
69 prod_hi = _mm256_srli_epi64(prod_hi, shift); in highbd_mul_shift_avx2()
|
| D | sum_squares_avx2.c | 51 v_acc_q = _mm256_add_epi64(v_acc_q, _mm256_srli_epi64(v_acc_d, 32)); in aom_sum_squares_2d_i16_nxn_avx2()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 3x8c8-minmax-avx2.c | 168 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
170 const __m256i vq31prod1x0246 = _mm256_srli_epi64(vprod1x0246, 31); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
172 const __m256i vq31prod2x0246 = _mm256_srli_epi64(vprod2x0246, 31); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
|
| D | 2x8c8-minmax-avx2.c | 142 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2()
144 const __m256i vq31prod1x0246 = _mm256_srli_epi64(vprod1x0246, 31); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2()
|
| D | 1x8c8-minmax-avx2.c | 116 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_igemm_minmax_ukernel_1x8c8__avx2()
|
| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 3x8c8-minmax-avx2.c | 151 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
153 const __m256i vq31prod1x0246 = _mm256_srli_epi64(vprod1x0246, 31); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
155 const __m256i vq31prod2x0246 = _mm256_srli_epi64(vprod2x0246, 31); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
|
| D | 3x8c8-xw-minmax-avx2.c | 147 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
149 const __m256i vq31prod1x0246 = _mm256_srli_epi64(vprod1x0246, 31); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
151 const __m256i vq31prod2x0246 = _mm256_srli_epi64(vprod2x0246, 31); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
|
| D | 2x8c8-minmax-avx2.c | 127 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2()
129 const __m256i vq31prod1x0246 = _mm256_srli_epi64(vprod1x0246, 31); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2()
|
| D | 2x8c8-xw-minmax-avx2.c | 123 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2()
125 const __m256i vq31prod1x0246 = _mm256_srli_epi64(vprod1x0246, 31); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2()
|
| D | 1x8c8-minmax-avx2.c | 103 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_gemm_minmax_ukernel_1x8c8__avx2()
|
| D | 1x8c8-xw-minmax-avx2.c | 99 const __m256i vq31prod0x0246 = _mm256_srli_epi64(vprod0x0246, 31); in xnn_qs8_gemm_xw_minmax_ukernel_1x8c8__avx2()
|
| /external/XNNPACK/src/qs8-dwconv/gen/ |
| D | up32x9-minmax-avx2-mul16.c | 243 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
245 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
247 const __m256i vq31prodGIKM = _mm256_srli_epi64(vprodGIKM, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
249 const __m256i vq31prodOQSU = _mm256_srli_epi64(vprodOQSU, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
396 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
398 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul16()
|
| D | up16x9-minmax-avx2-mul16.c | 181 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16()
183 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16()
304 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16()
306 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul16()
|
| D | up32x9-minmax-avx2-mul32.c | 243 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32()
245 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32()
247 const __m256i vq31prodGIKM = _mm256_srli_epi64(vprodGIKM, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32()
249 const __m256i vq31prodOQSU = _mm256_srli_epi64(vprodOQSU, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32()
365 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx2_mul32()
|
| D | up24x9-minmax-avx2-mul32.c | 212 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32()
214 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32()
216 const __m256i vq31prodGIKM = _mm256_srli_epi64(vprodGIKM, 31); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32()
327 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up24x9__avx2_mul32()
|
| D | up16x9-minmax-avx2-mul32.c | 181 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul32()
183 const __m256i vq31prod8ACE = _mm256_srli_epi64(vprod8ACE, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul32()
285 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up16x9__avx2_mul32()
|
| D | up8x9-minmax-avx2-mul32.c | 150 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up8x9__avx2_mul32()
235 const __m256i vq31prod0246 = _mm256_srli_epi64(vprod0246, 31); in xnn_qs8_dwconv_minmax_ukernel_up8x9__avx2_mul32()
|
| /external/XNNPACK/src/qs8-dwconv/ |
| D | unipass-avx2-mul16.c.in | 77 const __m256i vq31prod${ABC[C:C+8:2]} = _mm256_srli_epi64(vprod${ABC[C:C+8:2]}, 31);
151 const __m256i vq31prod${ABC[0:8:2]} = _mm256_srli_epi64(vprod${ABC[0:8:2]}, 31);
153 const __m256i vq31prod${ABC[8:16:2]} = _mm256_srli_epi64(vprod${ABC[8:16:2]}, 31);
|
| D | unipass-avx2-mul32.c.in | 74 const __m256i vq31prod${ABC[C:C+8:2]} = _mm256_srli_epi64(vprod${ABC[C:C+8:2]}, 31);
167 const __m256i vq31prod${ABC[0:8:2]} = _mm256_srli_epi64(vprod${ABC[0:8:2]}, 31);
|
| /external/libvpx/libvpx/vpx_dsp/x86/ |
| D | avg_intrin_avx2.c | 453 const __m256i c = _mm256_srli_epi64(b, 32); in vpx_satd_avx2()
475 const __m256i c = _mm256_srli_epi64(b, 32); in vpx_highbd_satd_avx2()
|