| /external/clang/lib/Headers/ |
| D | xopintrin.h | 36 static __inline__ __m128i __DEFAULT_FN_ATTRS
37 _mm_maccs_epi16(__m128i __A, __m128i __B, __m128i __C) in _mm_maccs_epi16()
39 return (__m128i)__builtin_ia32_vpmacssww((__v8hi)__A, (__v8hi)__B, (__v8hi)__C); in _mm_maccs_epi16()
42 static __inline__ __m128i __DEFAULT_FN_ATTRS
43 _mm_macc_epi16(__m128i __A, __m128i __B, __m128i __C) in _mm_macc_epi16()
45 return (__m128i)__builtin_ia32_vpmacsww((__v8hi)__A, (__v8hi)__B, (__v8hi)__C); in _mm_macc_epi16()
48 static __inline__ __m128i __DEFAULT_FN_ATTRS
49 _mm_maccsd_epi16(__m128i __A, __m128i __B, __m128i __C) in _mm_maccsd_epi16()
51 return (__m128i)__builtin_ia32_vpmacsswd((__v8hi)__A, (__v8hi)__B, (__v4si)__C); in _mm_maccsd_epi16()
54 static __inline__ __m128i __DEFAULT_FN_ATTRS
[all …]
|
| /external/clang/test/CodeGen/ |
| D | xop-builtins.c | 11 __m128i test_mm_maccs_epi16(__m128i a, __m128i b, __m128i c) { in test_mm_maccs_epi16()
17 __m128i test_mm_macc_epi16(__m128i a, __m128i b, __m128i c) { in test_mm_macc_epi16()
23 __m128i test_mm_maccsd_epi16(__m128i a, __m128i b, __m128i c) { in test_mm_maccsd_epi16()
29 __m128i test_mm_maccd_epi16(__m128i a, __m128i b, __m128i c) { in test_mm_maccd_epi16()
35 __m128i test_mm_maccs_epi32(__m128i a, __m128i b, __m128i c) { in test_mm_maccs_epi32()
41 __m128i test_mm_macc_epi32(__m128i a, __m128i b, __m128i c) { in test_mm_macc_epi32()
47 __m128i test_mm_maccslo_epi32(__m128i a, __m128i b, __m128i c) { in test_mm_maccslo_epi32()
53 __m128i test_mm_macclo_epi32(__m128i a, __m128i b, __m128i c) { in test_mm_macclo_epi32()
59 __m128i test_mm_maccshi_epi32(__m128i a, __m128i b, __m128i c) { in test_mm_maccshi_epi32()
65 __m128i test_mm_macchi_epi32(__m128i a, __m128i b, __m128i c) { in test_mm_macchi_epi32()
[all …]
|
| D | sse2-builtins.c | 11 __m128i test_mm_add_epi8(__m128i A, __m128i B) { in test_mm_add_epi8()
17 __m128i test_mm_add_epi16(__m128i A, __m128i B) { in test_mm_add_epi16()
23 __m128i test_mm_add_epi32(__m128i A, __m128i B) { in test_mm_add_epi32()
29 __m128i test_mm_add_epi64(__m128i A, __m128i B) { in test_mm_add_epi64()
50 __m128i test_mm_adds_epi8(__m128i A, __m128i B) { in test_mm_adds_epi8()
56 __m128i test_mm_adds_epi16(__m128i A, __m128i B) { in test_mm_adds_epi16()
62 __m128i test_mm_adds_epu8(__m128i A, __m128i B) { in test_mm_adds_epu8()
68 __m128i test_mm_adds_epu16(__m128i A, __m128i B) { in test_mm_adds_epu16()
80 __m128i test_mm_and_si128(__m128i A, __m128i B) { in test_mm_and_si128()
93 __m128i test_mm_andnot_si128(__m128i A, __m128i B) { in test_mm_andnot_si128()
[all …]
|
| D | sse41-builtins.c | 11 __m128i test_mm_blend_epi16(__m128i V1, __m128i V2) { in test_mm_blend_epi16()
29 __m128i test_mm_blendv_epi8(__m128i V1, __m128i V2, __m128i V3) { in test_mm_blendv_epi8()
71 __m128i test_mm_cmpeq_epi64(__m128i A, __m128i B) { in test_mm_cmpeq_epi64()
78 __m128i test_mm_cvtepi8_epi16(__m128i a) { in test_mm_cvtepi8_epi16()
85 __m128i test_mm_cvtepi8_epi32(__m128i a) { in test_mm_cvtepi8_epi32()
92 __m128i test_mm_cvtepi8_epi64(__m128i a) { in test_mm_cvtepi8_epi64()
99 __m128i test_mm_cvtepi16_epi32(__m128i a) { in test_mm_cvtepi16_epi32()
106 __m128i test_mm_cvtepi16_epi64(__m128i a) { in test_mm_cvtepi16_epi64()
113 __m128i test_mm_cvtepi32_epi64(__m128i a) { in test_mm_cvtepi32_epi64()
120 __m128i test_mm_cvtepu8_epi16(__m128i a) { in test_mm_cvtepu8_epi16()
[all …]
|
| D | avx512vlbw-builtins.c | 22 __mmask16 test_mm_cmpeq_epi8_mask(__m128i __a, __m128i __b) { in test_mm_cmpeq_epi8_mask()
28 __mmask16 test_mm_mask_cmpeq_epi8_mask(__mmask16 __u, __m128i __a, __m128i __b) { in test_mm_mask_cmpeq_epi8_mask()
48 __mmask8 test_mm_cmpeq_epi16_mask(__m128i __a, __m128i __b) { in test_mm_cmpeq_epi16_mask()
54 __mmask8 test_mm_mask_cmpeq_epi16_mask(__mmask8 __u, __m128i __a, __m128i __b) { in test_mm_mask_cmpeq_epi16_mask()
74 __mmask16 test_mm_cmpgt_epi8_mask(__m128i __a, __m128i __b) { in test_mm_cmpgt_epi8_mask()
80 __mmask16 test_mm_mask_cmpgt_epi8_mask(__mmask16 __u, __m128i __a, __m128i __b) { in test_mm_mask_cmpgt_epi8_mask()
100 __mmask8 test_mm_cmpgt_epi16_mask(__m128i __a, __m128i __b) { in test_mm_cmpgt_epi16_mask()
106 __mmask8 test_mm_mask_cmpgt_epi16_mask(__mmask8 __u, __m128i __a, __m128i __b) { in test_mm_mask_cmpgt_epi16_mask()
113 __mmask16 test_mm_cmpeq_epu8_mask(__m128i __a, __m128i __b) { in test_mm_cmpeq_epu8_mask()
119 __mmask16 test_mm_mask_cmpeq_epu8_mask(__mmask16 __u, __m128i __a, __m128i __b) { in test_mm_mask_cmpeq_epu8_mask()
[all …]
|
| /external/libvpx/libvpx/vpx_dsp/x86/ |
| D | fwd_txfm_sse2.h | 21 static INLINE __m128i k_madd_epi32(__m128i a, __m128i b) { in k_madd_epi32()
22 __m128i buf0, buf1; in k_madd_epi32()
30 static INLINE __m128i k_packs_epi64(__m128i a, __m128i b) { in k_packs_epi64()
31 __m128i buf0 = _mm_shuffle_epi32(a, _MM_SHUFFLE(0, 0, 2, 0)); in k_packs_epi64()
32 __m128i buf1 = _mm_shuffle_epi32(b, _MM_SHUFFLE(0, 0, 2, 0)); in k_packs_epi64()
36 static INLINE int check_epi16_overflow_x2(const __m128i *preg0, in check_epi16_overflow_x2()
37 const __m128i *preg1) { in check_epi16_overflow_x2()
38 const __m128i max_overflow = _mm_set1_epi16(0x7fff); in check_epi16_overflow_x2()
39 const __m128i min_overflow = _mm_set1_epi16(0x8000); in check_epi16_overflow_x2()
40 __m128i cmp0 = _mm_or_si128(_mm_cmpeq_epi16(*preg0, max_overflow), in check_epi16_overflow_x2()
[all …]
|
| D | highbd_intrapred_intrin_ssse3.c | 29 static INLINE __m128i avg3_epu16(const __m128i *x, const __m128i *y, in avg3_epu16()
30 const __m128i *z) { in avg3_epu16()
31 const __m128i one = _mm_set1_epi16(1); in avg3_epu16()
32 const __m128i a = _mm_avg_epu16(*x, *z); in avg3_epu16()
33 const __m128i b = in avg3_epu16()
41 const __m128i ABCDEFGH = _mm_loadu_si128((const __m128i *)above); in vpx_highbd_d45_predictor_4x4_ssse3()
42 const __m128i BCDEFGH0 = _mm_srli_si128(ABCDEFGH, 2); in vpx_highbd_d45_predictor_4x4_ssse3()
43 const __m128i CDEFGH00 = _mm_srli_si128(ABCDEFGH, 4); in vpx_highbd_d45_predictor_4x4_ssse3()
44 const __m128i avg3 = avg3_epu16(&ABCDEFGH, &BCDEFGH0, &CDEFGH00); in vpx_highbd_d45_predictor_4x4_ssse3()
47 _mm_storel_epi64((__m128i *)dst, avg3); in vpx_highbd_d45_predictor_4x4_ssse3()
[all …]
|
| D | highbd_intrapred_intrin_sse2.c | 22 const __m128i left_u16 = _mm_loadl_epi64((const __m128i *)left); in vpx_highbd_h_predictor_4x4_sse2()
23 const __m128i row0 = _mm_shufflelo_epi16(left_u16, 0x0); in vpx_highbd_h_predictor_4x4_sse2()
24 const __m128i row1 = _mm_shufflelo_epi16(left_u16, 0x55); in vpx_highbd_h_predictor_4x4_sse2()
25 const __m128i row2 = _mm_shufflelo_epi16(left_u16, 0xaa); in vpx_highbd_h_predictor_4x4_sse2()
26 const __m128i row3 = _mm_shufflelo_epi16(left_u16, 0xff); in vpx_highbd_h_predictor_4x4_sse2()
29 _mm_storel_epi64((__m128i *)dst, row0); in vpx_highbd_h_predictor_4x4_sse2()
31 _mm_storel_epi64((__m128i *)dst, row1); in vpx_highbd_h_predictor_4x4_sse2()
33 _mm_storel_epi64((__m128i *)dst, row2); in vpx_highbd_h_predictor_4x4_sse2()
35 _mm_storel_epi64((__m128i *)dst, row3); in vpx_highbd_h_predictor_4x4_sse2()
41 const __m128i left_u16 = _mm_load_si128((const __m128i *)left); in vpx_highbd_h_predictor_8x8_sse2()
[all …]
|
| /external/libaom/libaom/aom_dsp/x86/ |
| D | fwd_txfm_sse2.h | 19 static INLINE __m128i k_madd_epi32(__m128i a, __m128i b) { in k_madd_epi32()
20 __m128i buf0, buf1; in k_madd_epi32()
28 static INLINE __m128i k_packs_epi64(__m128i a, __m128i b) { in k_packs_epi64()
29 __m128i buf0 = _mm_shuffle_epi32(a, _MM_SHUFFLE(0, 0, 2, 0)); in k_packs_epi64()
30 __m128i buf1 = _mm_shuffle_epi32(b, _MM_SHUFFLE(0, 0, 2, 0)); in k_packs_epi64()
34 static INLINE int check_epi16_overflow_x2(const __m128i *preg0, in check_epi16_overflow_x2()
35 const __m128i *preg1) { in check_epi16_overflow_x2()
36 const __m128i max_overflow = _mm_set1_epi16(0x7fff); in check_epi16_overflow_x2()
37 const __m128i min_overflow = _mm_set1_epi16(0x8000); in check_epi16_overflow_x2()
38 __m128i cmp0 = _mm_or_si128(_mm_cmpeq_epi16(*preg0, max_overflow), in check_epi16_overflow_x2()
[all …]
|
| D | blend_sse4.h | 26 static INLINE __m128i blend_4(const uint8_t *src0, const uint8_t *src1, in blend_4()
27 const __m128i *v_m0_w, const __m128i *v_m1_w) { in blend_4()
28 const __m128i v_s0_b = xx_loadl_32(src0); in blend_4()
29 const __m128i v_s1_b = xx_loadl_32(src1); in blend_4()
30 const __m128i v_s0_w = _mm_cvtepu8_epi16(v_s0_b); in blend_4()
31 const __m128i v_s1_w = _mm_cvtepu8_epi16(v_s1_b); in blend_4()
33 const __m128i v_p0_w = _mm_mullo_epi16(v_s0_w, *v_m0_w); in blend_4()
34 const __m128i v_p1_w = _mm_mullo_epi16(v_s1_w, *v_m1_w); in blend_4()
35 const __m128i v_sum_w = _mm_add_epi16(v_p0_w, v_p1_w); in blend_4()
36 const __m128i v_res_w = xx_roundn_epu16(v_sum_w, AOM_BLEND_A64_ROUND_BITS); in blend_4()
[all …]
|
| D | lpf_common_sse2.h | 19 static INLINE void highbd_transpose6x6_sse2(__m128i *x0, __m128i *x1, in highbd_transpose6x6_sse2()
20 __m128i *x2, __m128i *x3, in highbd_transpose6x6_sse2()
21 __m128i *x4, __m128i *x5, in highbd_transpose6x6_sse2()
22 __m128i *d0, __m128i *d1, in highbd_transpose6x6_sse2()
23 __m128i *d2, __m128i *d3, in highbd_transpose6x6_sse2()
24 __m128i *d4, __m128i *d5) { in highbd_transpose6x6_sse2()
25 __m128i w0, w1, w2, w3, w4, w5, ww0; in highbd_transpose6x6_sse2()
59 static INLINE void highbd_transpose4x8_8x4_low_sse2(__m128i *x0, __m128i *x1, in highbd_transpose4x8_8x4_low_sse2()
60 __m128i *x2, __m128i *x3, in highbd_transpose4x8_8x4_low_sse2()
61 __m128i *d0, __m128i *d1, in highbd_transpose4x8_8x4_low_sse2()
[all …]
|
| D | highbd_intrapred_sse2.c | 22 const __m128i left_u16 = _mm_loadl_epi64((const __m128i *)left); in aom_highbd_h_predictor_4x4_sse2()
23 const __m128i row0 = _mm_shufflelo_epi16(left_u16, 0x0); in aom_highbd_h_predictor_4x4_sse2()
24 const __m128i row1 = _mm_shufflelo_epi16(left_u16, 0x55); in aom_highbd_h_predictor_4x4_sse2()
25 const __m128i row2 = _mm_shufflelo_epi16(left_u16, 0xaa); in aom_highbd_h_predictor_4x4_sse2()
26 const __m128i row3 = _mm_shufflelo_epi16(left_u16, 0xff); in aom_highbd_h_predictor_4x4_sse2()
29 _mm_storel_epi64((__m128i *)dst, row0); in aom_highbd_h_predictor_4x4_sse2()
31 _mm_storel_epi64((__m128i *)dst, row1); in aom_highbd_h_predictor_4x4_sse2()
33 _mm_storel_epi64((__m128i *)dst, row2); in aom_highbd_h_predictor_4x4_sse2()
35 _mm_storel_epi64((__m128i *)dst, row3); in aom_highbd_h_predictor_4x4_sse2()
50 const __m128i left_u16 = _mm_load_si128((const __m128i *)left); in aom_highbd_h_predictor_8x4_sse2()
[all …]
|
| D | intrapred_ssse3.c | 22 static INLINE __m128i paeth_8x1_pred(const __m128i *left, const __m128i *top, in paeth_8x1_pred()
23 const __m128i *topleft) { in paeth_8x1_pred()
24 const __m128i base = _mm_sub_epi16(_mm_add_epi16(*top, *left), *topleft); in paeth_8x1_pred()
26 __m128i pl = _mm_abs_epi16(_mm_sub_epi16(base, *left)); in paeth_8x1_pred()
27 __m128i pt = _mm_abs_epi16(_mm_sub_epi16(base, *top)); in paeth_8x1_pred()
28 __m128i ptl = _mm_abs_epi16(_mm_sub_epi16(base, *topleft)); in paeth_8x1_pred()
30 __m128i mask1 = _mm_cmpgt_epi16(pl, pt); in paeth_8x1_pred()
32 __m128i mask2 = _mm_cmpgt_epi16(pt, ptl); in paeth_8x1_pred()
46 __m128i l = _mm_loadl_epi64((const __m128i *)left); in aom_paeth_predictor_4x4_ssse3()
47 const __m128i t = _mm_loadl_epi64((const __m128i *)above); in aom_paeth_predictor_4x4_ssse3()
[all …]
|
| /external/libaom/libaom/av1/common/x86/ |
| D | cfl_ssse3.c | 21 static INLINE __m128i _mm_loadh_epi32(__m128i const *mem_addr) { in _mm_loadh_epi32()
26 static INLINE void _mm_storeh_epi32(__m128i const *mem_addr, __m128i a) { in _mm_storeh_epi32()
44 const __m128i twos = _mm_set1_epi8(2); in cfl_luma_subsampling_420_lbd_ssse3()
45 __m128i *pred_buf_m128i = (__m128i *)pred_buf_q3; in cfl_luma_subsampling_420_lbd_ssse3()
46 const __m128i *end = pred_buf_m128i + (height >> 1) * CFL_BUF_LINE_I128; in cfl_luma_subsampling_420_lbd_ssse3()
50 __m128i top = _mm_loadh_epi32((__m128i *)input); in cfl_luma_subsampling_420_lbd_ssse3()
52 __m128i bot = _mm_loadh_epi32((__m128i *)(input + input_stride)); in cfl_luma_subsampling_420_lbd_ssse3()
54 const __m128i sum = _mm_add_epi16(top, bot); in cfl_luma_subsampling_420_lbd_ssse3()
57 __m128i top = _mm_loadl_epi64((__m128i *)input); in cfl_luma_subsampling_420_lbd_ssse3()
59 __m128i bot = _mm_loadl_epi64((__m128i *)(input + input_stride)); in cfl_luma_subsampling_420_lbd_ssse3()
[all …]
|
| D | convolve_2d_sse2.c | 38 const __m128i zero = _mm_setzero_si128(); in av1_convolve_2d_sr_sse2()
49 const __m128i coeffs_x = _mm_loadu_si128((__m128i *)x_filter); in av1_convolve_2d_sr_sse2()
52 const __m128i tmp_0 = _mm_unpacklo_epi32(coeffs_x, coeffs_x); in av1_convolve_2d_sr_sse2()
54 const __m128i tmp_1 = _mm_unpackhi_epi32(coeffs_x, coeffs_x); in av1_convolve_2d_sr_sse2()
57 const __m128i coeff_01 = _mm_unpacklo_epi64(tmp_0, tmp_0); in av1_convolve_2d_sr_sse2()
59 const __m128i coeff_23 = _mm_unpackhi_epi64(tmp_0, tmp_0); in av1_convolve_2d_sr_sse2()
61 const __m128i coeff_45 = _mm_unpacklo_epi64(tmp_1, tmp_1); in av1_convolve_2d_sr_sse2()
63 const __m128i coeff_67 = _mm_unpackhi_epi64(tmp_1, tmp_1); in av1_convolve_2d_sr_sse2()
65 const __m128i round_const = _mm_set1_epi32( in av1_convolve_2d_sr_sse2()
67 const __m128i round_shift = _mm_cvtsi32_si128(conv_params->round_0); in av1_convolve_2d_sr_sse2()
[all …]
|
| D | highbd_convolve_2d_sse4.c | 38 const __m128i left_shift = _mm_cvtsi32_si128(bits); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
43 const __m128i wt0 = _mm_set1_epi32(w0); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
44 const __m128i wt1 = _mm_set1_epi32(w1); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
45 const __m128i zero = _mm_setzero_si128(); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
51 const __m128i offset_const = _mm_set1_epi32(offset); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
52 const __m128i offset_const_16b = _mm_set1_epi16(offset); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
55 const __m128i rounding_const = _mm_set1_epi32((1 << rounding_shift) >> 1); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
56 const __m128i clip_pixel_to_bd = in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
64 const __m128i src_16bit = in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
65 _mm_loadu_si128((__m128i *)(&src[i * src_stride + j])); in av1_highbd_dist_wtd_convolve_2d_copy_sse4_1()
[all …]
|
| D | intra_edge_sse4.c | 36 __m128i last = _mm_set1_epi8(p[sz - 1]); in av1_filter_intra_edge_sse4_1()
37 _mm_storeu_si128((__m128i *)&p[sz], last); in av1_filter_intra_edge_sse4_1()
49 __m128i coef0 = _mm_lddqu_si128((__m128i const *)kern[strength - 1]); in av1_filter_intra_edge_sse4_1()
50 __m128i shuf0 = _mm_lddqu_si128((__m128i const *)v_const[0]); in av1_filter_intra_edge_sse4_1()
51 __m128i shuf1 = _mm_lddqu_si128((__m128i const *)v_const[1]); in av1_filter_intra_edge_sse4_1()
52 __m128i iden = _mm_lddqu_si128((__m128i *)v_const[3]); in av1_filter_intra_edge_sse4_1()
53 __m128i in0 = _mm_lddqu_si128((__m128i *)in); in av1_filter_intra_edge_sse4_1()
56 __m128i d0 = _mm_shuffle_epi8(in0, shuf0); in av1_filter_intra_edge_sse4_1()
57 __m128i d1 = _mm_shuffle_epi8(in0, shuf1); in av1_filter_intra_edge_sse4_1()
61 __m128i eight = _mm_set1_epi16(8); in av1_filter_intra_edge_sse4_1()
[all …]
|
| D | highbd_convolve_2d_sse2.c | 19 __m128i s[8]; in copy_64()
20 s[0] = _mm_loadu_si128((__m128i *)(src + 0 * 8)); in copy_64()
21 s[1] = _mm_loadu_si128((__m128i *)(src + 1 * 8)); in copy_64()
22 s[2] = _mm_loadu_si128((__m128i *)(src + 2 * 8)); in copy_64()
23 s[3] = _mm_loadu_si128((__m128i *)(src + 3 * 8)); in copy_64()
24 s[4] = _mm_loadu_si128((__m128i *)(src + 4 * 8)); in copy_64()
25 s[5] = _mm_loadu_si128((__m128i *)(src + 5 * 8)); in copy_64()
26 s[6] = _mm_loadu_si128((__m128i *)(src + 6 * 8)); in copy_64()
27 s[7] = _mm_loadu_si128((__m128i *)(src + 7 * 8)); in copy_64()
28 _mm_store_si128((__m128i *)(dst + 0 * 8), s[0]); in copy_64()
[all …]
|
| D | jnt_convolve_ssse3.c | 39 const __m128i zero = _mm_setzero_si128(); in av1_dist_wtd_convolve_2d_ssse3()
43 const __m128i wt0 = _mm_set1_epi16(w0); in av1_dist_wtd_convolve_2d_ssse3()
44 const __m128i wt1 = _mm_set1_epi16(w1); in av1_dist_wtd_convolve_2d_ssse3()
45 const __m128i wt = _mm_unpacklo_epi16(wt0, wt1); in av1_dist_wtd_convolve_2d_ssse3()
50 const __m128i offset_const = _mm_set1_epi16(offset); in av1_dist_wtd_convolve_2d_ssse3()
53 const __m128i rounding_const = _mm_set1_epi16((1 << rounding_shift) >> 1); in av1_dist_wtd_convolve_2d_ssse3()
59 const __m128i coeffs_x = _mm_loadu_si128((__m128i *)x_filter); in av1_dist_wtd_convolve_2d_ssse3()
62 const __m128i tmp_0 = _mm_unpacklo_epi32(coeffs_x, coeffs_x); in av1_dist_wtd_convolve_2d_ssse3()
64 const __m128i tmp_1 = _mm_unpackhi_epi32(coeffs_x, coeffs_x); in av1_dist_wtd_convolve_2d_ssse3()
67 const __m128i coeff_01 = _mm_unpacklo_epi64(tmp_0, tmp_0); in av1_dist_wtd_convolve_2d_ssse3()
[all …]
|
| /external/webp/src/dsp/ |
| D | lossless_enc_sse2.c | 33 const __m128i in = _mm_loadu_si128((__m128i*)&argb_data[i]); // argb in SubtractGreenFromBlueAndRed_SSE2()
34 const __m128i A = _mm_srli_epi16(in, 8); // 0 a 0 g in SubtractGreenFromBlueAndRed_SSE2()
35 const __m128i B = _mm_shufflelo_epi16(A, _MM_SHUFFLE(2, 2, 0, 0)); in SubtractGreenFromBlueAndRed_SSE2()
36 const __m128i C = _mm_shufflehi_epi16(B, _MM_SHUFFLE(2, 2, 0, 0)); // 0g0g in SubtractGreenFromBlueAndRed_SSE2()
37 const __m128i out = _mm_sub_epi8(in, C); in SubtractGreenFromBlueAndRed_SSE2()
38 _mm_storeu_si128((__m128i*)&argb_data[i], out); in SubtractGreenFromBlueAndRed_SSE2()
54 const __m128i mults_rb = MK_CST_16(CST_5b(m->green_to_red_), in TransformColor_SSE2()
56 const __m128i mults_b2 = MK_CST_16(CST_5b(m->red_to_blue_), 0); in TransformColor_SSE2()
57 const __m128i mask_ag = _mm_set1_epi32(0xff00ff00); // alpha-green masks in TransformColor_SSE2()
58 const __m128i mask_rb = _mm_set1_epi32(0x00ff00ff); // red-blue masks in TransformColor_SSE2()
[all …]
|
| D | enc_sse2.c | 47 const __m128i k1 = _mm_set1_epi16(20091); in ITransform_SSE2()
48 const __m128i k2 = _mm_set1_epi16(-30068); in ITransform_SSE2()
49 __m128i T0, T1, T2, T3; in ITransform_SSE2()
55 __m128i in0, in1, in2, in3; in ITransform_SSE2()
57 in0 = _mm_loadl_epi64((const __m128i*)&in[0]); in ITransform_SSE2()
58 in1 = _mm_loadl_epi64((const __m128i*)&in[4]); in ITransform_SSE2()
59 in2 = _mm_loadl_epi64((const __m128i*)&in[8]); in ITransform_SSE2()
60 in3 = _mm_loadl_epi64((const __m128i*)&in[12]); in ITransform_SSE2()
66 const __m128i inB0 = _mm_loadl_epi64((const __m128i*)&in[16]); in ITransform_SSE2()
67 const __m128i inB1 = _mm_loadl_epi64((const __m128i*)&in[20]); in ITransform_SSE2()
[all …]
|
| D | lossless_sse2.c | 30 const __m128i zero = _mm_setzero_si128(); in ClampedAddSubtractFull_SSE2()
31 const __m128i C0 = _mm_unpacklo_epi8(_mm_cvtsi32_si128(c0), zero); in ClampedAddSubtractFull_SSE2()
32 const __m128i C1 = _mm_unpacklo_epi8(_mm_cvtsi32_si128(c1), zero); in ClampedAddSubtractFull_SSE2()
33 const __m128i C2 = _mm_unpacklo_epi8(_mm_cvtsi32_si128(c2), zero); in ClampedAddSubtractFull_SSE2()
34 const __m128i V1 = _mm_add_epi16(C0, C1); in ClampedAddSubtractFull_SSE2()
35 const __m128i V2 = _mm_sub_epi16(V1, C2); in ClampedAddSubtractFull_SSE2()
36 const __m128i b = _mm_packus_epi16(V2, V2); in ClampedAddSubtractFull_SSE2()
44 const __m128i zero = _mm_setzero_si128(); in ClampedAddSubtractHalf_SSE2()
45 const __m128i C0 = _mm_unpacklo_epi8(_mm_cvtsi32_si128(c0), zero); in ClampedAddSubtractHalf_SSE2()
46 const __m128i C1 = _mm_unpacklo_epi8(_mm_cvtsi32_si128(c1), zero); in ClampedAddSubtractHalf_SSE2()
[all …]
|
| /external/libmpeg2/common/x86/ |
| D | impeg2_inter_pred_sse42_intr.c | 76 __m128i src_r0, src_r1, src_r2, src_r3; in impeg2_copy_mb_sse42()
84 src_r0 = _mm_loadu_si128((__m128i *) (src)); in impeg2_copy_mb_sse42()
85 src_r1 = _mm_loadu_si128((__m128i *) (src + src_wd)); in impeg2_copy_mb_sse42()
86 src_r2 = _mm_loadu_si128((__m128i *) (src + 2 * src_wd)); in impeg2_copy_mb_sse42()
87 src_r3 = _mm_loadu_si128((__m128i *) (src + 3 * src_wd)); in impeg2_copy_mb_sse42()
89 _mm_storeu_si128((__m128i *) dst, src_r0); in impeg2_copy_mb_sse42()
90 _mm_storeu_si128((__m128i *) (dst + dst_wd), src_r1); in impeg2_copy_mb_sse42()
91 _mm_storeu_si128((__m128i *) (dst + 2 * dst_wd), src_r2); in impeg2_copy_mb_sse42()
92 _mm_storeu_si128((__m128i *) (dst + 3 * dst_wd), src_r3); in impeg2_copy_mb_sse42()
97 src_r0 = _mm_loadu_si128((__m128i *) (src)); in impeg2_copy_mb_sse42()
[all …]
|
| /external/libaom/libaom/av1/encoder/x86/ |
| D | wedge_utils_sse2.c | 33 const __m128i v_mask_max_w = _mm_set1_epi16(MAX_MASK_VALUE); in av1_wedge_sse_from_residuals_sse2()
34 const __m128i v_zext_q = xx_set1_64_from_32i(0xffffffff); in av1_wedge_sse_from_residuals_sse2()
36 __m128i v_acc0_q = _mm_setzero_si128(); in av1_wedge_sse_from_residuals_sse2()
45 const __m128i v_r0_w = xx_load_128(r1 + n); in av1_wedge_sse_from_residuals_sse2()
46 const __m128i v_r1_w = xx_load_128(r1 + n8); in av1_wedge_sse_from_residuals_sse2()
47 const __m128i v_d0_w = xx_load_128(d + n); in av1_wedge_sse_from_residuals_sse2()
48 const __m128i v_d1_w = xx_load_128(d + n8); in av1_wedge_sse_from_residuals_sse2()
49 const __m128i v_m01_b = xx_load_128(m + n); in av1_wedge_sse_from_residuals_sse2()
51 const __m128i v_rd0l_w = _mm_unpacklo_epi16(v_d0_w, v_r0_w); in av1_wedge_sse_from_residuals_sse2()
52 const __m128i v_rd0h_w = _mm_unpackhi_epi16(v_d0_w, v_r0_w); in av1_wedge_sse_from_residuals_sse2()
[all …]
|
| /external/libvpx/libvpx/vp8/common/x86/ |
| D | bilinear_filter_sse2.c | 23 const __m128i zero = _mm_setzero_si128(); in horizontal_16x16()
27 const __m128i a = _mm_loadu_si128((__m128i *)src); in horizontal_16x16()
28 const __m128i a_lo = _mm_unpacklo_epi8(a, zero); in horizontal_16x16()
29 const __m128i a_hi = _mm_unpackhi_epi8(a, zero); in horizontal_16x16()
30 _mm_store_si128((__m128i *)dst, a_lo); in horizontal_16x16()
31 _mm_store_si128((__m128i *)(dst + 8), a_hi); in horizontal_16x16()
39 const __m128i round_factor = _mm_set1_epi16(1 << (VP8_FILTER_SHIFT - 1)); in horizontal_16x16()
40 const __m128i hfilter_0 = _mm_set1_epi16(vp8_bilinear_filters[xoffset][0]); in horizontal_16x16()
41 const __m128i hfilter_1 = _mm_set1_epi16(vp8_bilinear_filters[xoffset][1]); in horizontal_16x16()
44 const __m128i a = _mm_loadu_si128((__m128i *)src); in horizontal_16x16()
[all …]
|