virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
Code Issues Releases Activity

avcodec/loongarch: Add wrapper for __lsx_vldx

Browse Source 
__lsx_vldx does not accept a pointer to const (in fact,
no function in lsxintrin.h does so), although it is not allowed
to modify the pointed-to buffer. Therefore this commit adds a wrapper
for it in order to constify the HEVC DSP functions in a later commit.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This commit is contained in:
Andreas Rheinhardt 2022-08-02 00:56:53 +02:00
parent 101e154c28
commit 6c9a60ada4
6 changed files with 252 additions and 247 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
libavcodec/loongarch/hevc_lpf_sao_lsx.c
View File

@ -1201,17 +1201,17 @@ static void hevc_sao_edge_filter_0degree_16multiple_lsx(uint8_t *dst,
for (; height; height -= 4) { for (; height; height -= 4) {
src_minus1 = src - 1; src_minus1 = src - 1;
src_minus10 = __lsx_vld(src_minus1, 0); src_minus10 = __lsx_vld(src_minus1, 0);
DUP2_ARG2(__lsx_vldx, src_minus1, src_stride, src_minus1, DUP2_ARG2(LSX_VLDX, src_minus1, src_stride, src_minus1,
src_stride_2x, src_minus11, src_minus12); src_stride_2x, src_minus11, src_minus12);
src_minus13 = __lsx_vldx(src_minus1, src_stride_3x); src_minus13 = LSX_VLDX(src_minus1, src_stride_3x);
for (v_cnt = 0; v_cnt < width; v_cnt += 16) { for (v_cnt = 0; v_cnt < width; v_cnt += 16) {
src_minus1 += 16; src_minus1 += 16;
dst_ptr = dst + v_cnt; dst_ptr = dst + v_cnt;
src10 = __lsx_vld(src_minus1, 0); src10 = __lsx_vld(src_minus1, 0);
DUP2_ARG2(__lsx_vldx, src_minus1, src_stride, src_minus1, DUP2_ARG2(LSX_VLDX, src_minus1, src_stride, src_minus1,
src_stride_2x, src11, src12); src_stride_2x, src11, src12);
src13 = __lsx_vldx(src_minus1, src_stride_3x); src13 = LSX_VLDX(src_minus1, src_stride_3x);
DUP4_ARG3(__lsx_vshuf_b, src10, src_minus10, shuf1, src11, DUP4_ARG3(__lsx_vshuf_b, src10, src_minus10, shuf1, src11,
src_minus11, shuf1, src12, src_minus12, shuf1, src13, src_minus11, shuf1, src12, src_minus12, shuf1, src13,
src_minus13, shuf1, src_zero0, src_zero1, src_minus13, shuf1, src_zero0, src_zero1,
@ -1358,7 +1358,7 @@ static void hevc_sao_edge_filter_90degree_4width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src10, src11); src10, src11);
__lsx_vstelm_w(dst0, dst, 0, 0); __lsx_vstelm_w(dst0, dst, 0, 0);
@ -1417,7 +1417,7 @@ static void hevc_sao_edge_filter_90degree_8width_lsx(uint8_t *dst,
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vld, src - src_stride, 0, src, 0, src_minus10, src_minus11); DUP2_ARG2(__lsx_vld, src - src_stride, 0, src, 0, src_minus10, src_minus11);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src10, src11); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src10, src11);
for (height -= 2; height; height -= 2) { for (height -= 2; height; height -= 2) {
src += src_stride_2x; src += src_stride_2x;
@ -1451,7 +1451,7 @@ static void hevc_sao_edge_filter_90degree_8width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src10, src11); src10, src11);
__lsx_vstelm_d(dst0, dst, 0, 0); __lsx_vstelm_d(dst0, dst, 0, 0);
@ -1528,7 +1528,7 @@ static void hevc_sao_edge_filter_90degree_16multiple_lsx(uint8_t *dst,
src_minus10, src_minus11); src_minus10, src_minus11);
for (h_cnt = (height >> 2); h_cnt--;) { for (h_cnt = (height >> 2); h_cnt--;) {
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src, src_stride_3x, src, src_stride_4x, src, src_stride_3x, src, src_stride_4x,
src10, src11, src12, src13); src10, src11, src12, src13);
DUP4_ARG2(__lsx_vseq_b, src_minus11, src_minus10, src_minus11, DUP4_ARG2(__lsx_vseq_b, src_minus11, src_minus10, src_minus11,
@ -1635,7 +1635,7 @@ static void hevc_sao_edge_filter_45degree_4width_lsx(uint8_t *dst,
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0, DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0,
src_minus10, src_minus11); src_minus10, src_minus11);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
for (height -= 2; height; height -= 2) { for (height -= 2; height; height -= 2) {
@ -1677,7 +1677,7 @@ static void hevc_sao_edge_filter_45degree_4width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
__lsx_vstelm_w(dst0, dst, 0, 0); __lsx_vstelm_w(dst0, dst, 0, 0);
@ -1748,7 +1748,7 @@ static void hevc_sao_edge_filter_45degree_8width_lsx(uint8_t *dst,
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0, src_minus10, DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0, src_minus10,
src_minus11); src_minus11);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
for (height -= 2; height; height -= 2) { for (height -= 2; height; height -= 2) {
@ -1790,7 +1790,7 @@ static void hevc_sao_edge_filter_45degree_8width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11) src10, src11)
__lsx_vstelm_d(dst0, dst, 0, 0); __lsx_vstelm_d(dst0, dst, 0, 0);
__lsx_vstelm_d(dst0, dst + dst_stride, 0, 1); __lsx_vstelm_d(dst0, dst + dst_stride, 0, 1);
@ -1833,7 +1833,7 @@ static void hevc_sao_edge_filter_45degree_8width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
__lsx_vstelm_d(dst0, dst, 0, 0); __lsx_vstelm_d(dst0, dst, 0, 0);
@ -1880,17 +1880,17 @@ static void hevc_sao_edge_filter_45degree_16multiple_lsx(uint8_t *dst,
src_orig = src - 1; src_orig = src - 1;
dst_orig = dst; dst_orig = dst;
src_minus11 = __lsx_vld(src_orig, 0); src_minus11 = __lsx_vld(src_orig, 0);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src_minus12, src_minus13); src_minus12, src_minus13);
src_minus14 = __lsx_vldx(src_orig, src_stride_3x); src_minus14 = LSX_VLDX(src_orig, src_stride_3x);
for (v_cnt = 0; v_cnt < width; v_cnt += 16) { for (v_cnt = 0; v_cnt < width; v_cnt += 16) {
src_minus10 = __lsx_vld(src_orig - src_stride, 0); src_minus10 = __lsx_vld(src_orig - src_stride, 0);
src_orig += 16; src_orig += 16;
src10 = __lsx_vld(src_orig, 0); src10 = __lsx_vld(src_orig, 0);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig,
src_stride_2x, src11, src12); src_stride_2x, src11, src12);
src13 = __lsx_vldx(src_orig, src_stride_3x); src13 = LSX_VLDX(src_orig, src_stride_3x);
src_plus13 = __lsx_vld(src + v_cnt + src_stride_4x, 1); src_plus13 = __lsx_vld(src + v_cnt + src_stride_4x, 1);
DUP4_ARG3(__lsx_vshuf_b, src10, src_minus11, shuf1, src11, DUP4_ARG3(__lsx_vshuf_b, src10, src_minus11, shuf1, src11,
@ -2016,7 +2016,7 @@ static void hevc_sao_edge_filter_135degree_4width_lsx(uint8_t *dst,
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0, DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0,
src_minus10, src_minus11); src_minus10, src_minus11);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
for (height -= 2; height; height -= 2) { for (height -= 2; height; height -= 2) {
@ -2058,7 +2058,7 @@ static void hevc_sao_edge_filter_135degree_4width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
__lsx_vstelm_w(dst0, dst, 0, 0); __lsx_vstelm_w(dst0, dst, 0, 0);
@ -2131,7 +2131,7 @@ static void hevc_sao_edge_filter_135degree_8width_lsx(uint8_t *dst,
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0, DUP2_ARG2(__lsx_vld, src_orig - src_stride, 0, src_orig, 0,
src_minus10, src_minus11); src_minus10, src_minus11);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
for (height -= 2; height; height -= 2) { for (height -= 2; height; height -= 2) {
@ -2173,7 +2173,7 @@ static void hevc_sao_edge_filter_135degree_8width_lsx(uint8_t *dst,
src_minus11 = src11; src_minus11 = src11;
/* load in advance */ /* load in advance */
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src10, src11); src10, src11);
__lsx_vstelm_d(dst0, dst, 0, 0); __lsx_vstelm_d(dst0, dst, 0, 0);
@ -2255,18 +2255,18 @@ static void hevc_sao_edge_filter_135degree_16multiple_lsx(uint8_t *dst,
dst_orig = dst; dst_orig = dst;
src_minus11 = __lsx_vld(src_orig, 0); src_minus11 = __lsx_vld(src_orig, 0);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src_plus10, src_plus11); src_plus10, src_plus11);
src_plus12 = __lsx_vldx(src_orig, src_stride_3x); src_plus12 = LSX_VLDX(src_orig, src_stride_3x);
for (v_cnt = 0; v_cnt < width; v_cnt += 16) { for (v_cnt = 0; v_cnt < width; v_cnt += 16) {
src_minus10 = __lsx_vld(src_orig - src_stride, 2); src_minus10 = __lsx_vld(src_orig - src_stride, 2);
src_plus13 = __lsx_vldx(src_orig, src_stride_4x); src_plus13 = LSX_VLDX(src_orig, src_stride_4x);
src_orig += 16; src_orig += 16;
src10 = __lsx_vld(src_orig, 0); src10 = __lsx_vld(src_orig, 0);
DUP2_ARG2(__lsx_vldx, src_orig, src_stride, src_orig, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_orig, src_stride, src_orig, src_stride_2x,
src11, src12); src11, src12);
src13 =__lsx_vldx(src_orig, src_stride_3x); src13 =LSX_VLDX(src_orig, src_stride_3x);
DUP4_ARG3(__lsx_vshuf_b, src10, src_minus11, shuf1, src11, DUP4_ARG3(__lsx_vshuf_b, src10, src_minus11, shuf1, src11,
src_plus10, shuf1, src12, src_plus11, shuf1, src13, src_plus10, shuf1, src12, src_plus11, shuf1, src13,

140
libavcodec/loongarch/hevc_mc_bi_lsx.c
View File

@ -163,14 +163,14 @@ void hevc_bi_copy_6w_lsx(uint8_t *src0_ptr, int32_t src_stride,
DUP2_ARG2(__lsx_vilvl_d, reg1, reg0, reg3, reg2, src2, src3); DUP2_ARG2(__lsx_vilvl_d, reg1, reg0, reg3, reg2, src2, src3);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
in4 = __lsx_vld(src1_ptr, 0); in4 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in5, in6); src2_stride_2x, in5, in6);
in7 = __lsx_vldx(src1_ptr, src2_stride_3x); in7 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6, DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6,
dst0, dst2, dst4, dst6); dst0, dst2, dst4, dst6);
@ -207,7 +207,7 @@ void hevc_bi_copy_6w_lsx(uint8_t *src0_ptr, int32_t src_stride,
src0 = __lsx_vilvl_d(reg1, reg0); src0 = __lsx_vilvl_d(reg1, reg0);
src0_ptr += src_stride_2x; src0_ptr += src_stride_2x;
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
in1 = __lsx_vldx(src1_ptr, src2_stride_x); in1 = LSX_VLDX(src1_ptr, src2_stride_x);
src1_ptr += src2_stride_x; src1_ptr += src2_stride_x;
dst0 = __lsx_vsllwil_hu_bu(src0, 6); dst0 = __lsx_vsllwil_hu_bu(src0, 6);
dst1 = __lsx_vilvh_b(zero, src0); dst1 = __lsx_vilvh_b(zero, src0);
@ -265,14 +265,14 @@ void hevc_bi_copy_8w_lsx(uint8_t *src0_ptr, int32_t src_stride,
DUP4_ARG2(__lsx_vslli_h, dst1, 6, dst3, 6, dst5, 6, dst7, 6, dst1, DUP4_ARG2(__lsx_vslli_h, dst1, 6, dst3, 6, dst5, 6, dst7, 6, dst1,
dst3, dst5, dst7); dst3, dst5, dst7);
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
in4 = __lsx_vld(src1_ptr, 0); in4 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in5, in6); src2_stride_2x, in5, in6);
in7 = __lsx_vldx(src1_ptr, src2_stride_3x); in7 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
out0 = hevc_bi_rnd_clip(in0, dst0, in1, dst1); out0 = hevc_bi_rnd_clip(in0, dst0, in1, dst1);
out1 = hevc_bi_rnd_clip(in2, dst2, in3, dst3); out1 = hevc_bi_rnd_clip(in2, dst2, in3, dst3);
@ -294,7 +294,7 @@ void hevc_bi_copy_8w_lsx(uint8_t *src0_ptr, int32_t src_stride,
reg1 = __lsx_vldrepl_d(src0_ptr + src_stride, 0); reg1 = __lsx_vldrepl_d(src0_ptr + src_stride, 0);
src0 = __lsx_vilvl_d(reg1, reg0); src0 = __lsx_vilvl_d(reg1, reg0);
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
in1 = __lsx_vldx(src1_ptr, src2_stride_x); in1 = LSX_VLDX(src1_ptr, src2_stride_x);
dst0 = __lsx_vsllwil_hu_bu(src0, 6); dst0 = __lsx_vsllwil_hu_bu(src0, 6);
dst1 = __lsx_vilvh_b(zero, src0); dst1 = __lsx_vilvh_b(zero, src0);
dst1 = __lsx_vslli_h(dst1, 6); dst1 = __lsx_vslli_h(dst1, 6);
@ -330,19 +330,19 @@ void hevc_bi_copy_12w_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (loop_cnt = 4; loop_cnt--;) { for (loop_cnt = 4; loop_cnt--;) {
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src0_ptr, src_stride_3x); src3 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
in4 = __lsx_vld(_src1, 0); in4 = __lsx_vld(_src1, 0);
DUP2_ARG2(__lsx_vldx, _src1, src2_stride_x, _src1, src2_stride_2x, DUP2_ARG2(LSX_VLDX, _src1, src2_stride_x, _src1, src2_stride_2x,
in5, in6); in5, in6);
in7 = __lsx_vldx(_src1, src2_stride_3x); in7 = LSX_VLDX(_src1, src2_stride_3x);
_src1 += src2_stride_2x; _src1 += src2_stride_2x;
DUP2_ARG2(__lsx_vilvl_d, in5, in4, in7, in6, in4, in5); DUP2_ARG2(__lsx_vilvl_d, in5, in4, in7, in6, in4, in5);
@ -389,19 +389,19 @@ void hevc_bi_copy_16w_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src0_ptr, src_stride_3x); src3 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
in4 = __lsx_vld(_src1, 0); in4 = __lsx_vld(_src1, 0);
DUP2_ARG2(__lsx_vldx, _src1, src2_stride_x, _src1, src2_stride_2x, DUP2_ARG2(LSX_VLDX, _src1, src2_stride_x, _src1, src2_stride_2x,
in5, in6); in5, in6);
in7 = __lsx_vldx(_src1, src2_stride_3x); in7 = LSX_VLDX(_src1, src2_stride_3x);
_src1 += src2_stride_2x; _src1 += src2_stride_2x;
DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6, DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6,
dst0_r, dst1_r, dst2_r, dst3_r) dst0_r, dst1_r, dst2_r, dst3_r)
@ -647,12 +647,12 @@ void hevc_vt_8t_8w_lsx(uint8_t *src0_ptr, int32_t src_stride, int16_t *src1_ptr,
filt0, filt1, filt2, filt3); filt0, filt1, filt2, filt3);
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src0_ptr, src_stride_3x); src3 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
src4 = __lsx_vld(src0_ptr, 0); src4 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src5, src6); src5, src6);
src0_ptr += src_stride_3x; src0_ptr += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
@ -661,14 +661,14 @@ void hevc_vt_8t_8w_lsx(uint8_t *src0_ptr, int32_t src_stride, int16_t *src1_ptr,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src7 = __lsx_vld(src0_ptr, 0); src7 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src8, src9); src8, src9);
src10 = __lsx_vldx(src0_ptr, src_stride_3x); src10 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, src2_stride_2x, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr, src2_stride_2x,
in1, in2); in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, src9, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, src9,
src76_r, src87_r, src98_r, src109_r); src76_r, src87_r, src98_r, src109_r);
@ -741,12 +741,12 @@ void hevc_vt_8t_16multx2mult_lsx(uint8_t *src0_ptr, int32_t src_stride,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src0_ptr_tmp, 0); src0 = __lsx_vld(src0_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr_tmp, src_stride, src0_ptr_tmp, DUP2_ARG2(LSX_VLDX, src0_ptr_tmp, src_stride, src0_ptr_tmp,
src_stride_2x, src1, src2); src_stride_2x, src1, src2);
src3 = __lsx_vldx(src0_ptr_tmp, src_stride_3x); src3 = LSX_VLDX(src0_ptr_tmp, src_stride_3x);
src0_ptr_tmp += src_stride_4x; src0_ptr_tmp += src_stride_4x;
src4 = __lsx_vld(src0_ptr_tmp, 0); src4 = __lsx_vld(src0_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr_tmp, src_stride, src0_ptr_tmp, DUP2_ARG2(LSX_VLDX, src0_ptr_tmp, src_stride, src0_ptr_tmp,
src_stride_2x, src5, src6); src_stride_2x, src5, src6);
src0_ptr_tmp += src_stride_3x; src0_ptr_tmp += src_stride_3x;
@ -759,7 +759,7 @@ void hevc_vt_8t_16multx2mult_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (loop_cnt = (height >> 1); loop_cnt--;) { for (loop_cnt = (height >> 1); loop_cnt--;) {
src7 = __lsx_vld(src0_ptr_tmp, 0); src7 = __lsx_vld(src0_ptr_tmp, 0);
src8 = __lsx_vldx(src0_ptr_tmp, src_stride); src8 = LSX_VLDX(src0_ptr_tmp, src_stride);
src0_ptr_tmp += src_stride_2x; src0_ptr_tmp += src_stride_2x;
DUP2_ARG2(__lsx_vld, src1_ptr_tmp, 0, src1_ptr_tmp, 16, in0, in2); DUP2_ARG2(__lsx_vld, src1_ptr_tmp, 0, src1_ptr_tmp, 16, in0, in2);
src1_ptr_tmp += src2_stride; src1_ptr_tmp += src2_stride;
@ -903,12 +903,12 @@ void hevc_hv_8t_8multx1mult_lsx(uint8_t *src0_ptr, int32_t src_stride,
src1_ptr_tmp = src1_ptr; src1_ptr_tmp = src1_ptr;
src0 = __lsx_vld(src0_ptr_tmp, 0); src0 = __lsx_vld(src0_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr_tmp, src_stride, src0_ptr_tmp, DUP2_ARG2(LSX_VLDX, src0_ptr_tmp, src_stride, src0_ptr_tmp,
src_stride_2x, src1, src2); src_stride_2x, src1, src2);
src3 = __lsx_vldx(src0_ptr_tmp, src_stride_3x); src3 = LSX_VLDX(src0_ptr_tmp, src_stride_3x);
src0_ptr_tmp += src_stride_4x; src0_ptr_tmp += src_stride_4x;
src4 = __lsx_vld(src0_ptr_tmp, 0); src4 = __lsx_vld(src0_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr_tmp, src_stride, src0_ptr_tmp, DUP2_ARG2(LSX_VLDX, src0_ptr_tmp, src_stride, src0_ptr_tmp,
src_stride_2x, src5, src6); src_stride_2x, src5, src6);
src0_ptr_tmp += src_stride_3x; src0_ptr_tmp += src_stride_3x;
@ -1134,9 +1134,9 @@ static void hevc_hz_4t_24w_lsx(uint8_t *src0_ptr, int32_t src_stride,
dst += dst_stride_4x; dst += dst_stride_4x;
in0 = __lsx_vld(src1_ptr_tmp, 0); in0 = __lsx_vld(src1_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr_tmp, src2_stride_x, src1_ptr_tmp, DUP2_ARG2(LSX_VLDX, src1_ptr_tmp, src2_stride_x, src1_ptr_tmp,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr_tmp, src2_stride_3x); in3 = LSX_VLDX(src1_ptr_tmp, src2_stride_3x);
src1_ptr_tmp += src2_stride_2x; src1_ptr_tmp += src2_stride_2x;
DUP4_ARG3(__lsx_vshuf_b, src1, src1, mask0, src3, src3, mask0, src5, DUP4_ARG3(__lsx_vshuf_b, src1, src1, mask0, src3, src3, mask0, src5,
@ -1229,7 +1229,7 @@ static void hevc_vt_4t_12w_lsx(uint8_t *src0_ptr, int32_t src_stride,
DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1); DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1);
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src0_ptr += src_stride_3x; src0_ptr += src_stride_3x;
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
@ -1238,19 +1238,19 @@ static void hevc_vt_4t_12w_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src3 = __lsx_vld(src0_ptr, 0); src3 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src0_ptr, src_stride_3x); src6 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
in4 = __lsx_vld(_src1, 0); in4 = __lsx_vld(_src1, 0);
DUP2_ARG2(__lsx_vldx, _src1, src2_stride_x, _src1, src2_stride_2x, DUP2_ARG2(LSX_VLDX, _src1, src2_stride_x, _src1, src2_stride_2x,
in5, in6); in5, in6);
in7 = __lsx_vldx(_src1, src2_stride_3x); in7 = LSX_VLDX(_src1, src2_stride_3x);
_src1 += src2_stride_2x; _src1 += src2_stride_2x;
DUP2_ARG2(__lsx_vilvl_d, in5, in4, in7, in6, in4, in5); DUP2_ARG2(__lsx_vilvl_d, in5, in4, in7, in6, in4, in5);
@ -1310,7 +1310,7 @@ static void hevc_vt_4t_16w_lsx(uint8_t *src0_ptr, int32_t src_stride,
DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1); DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1);
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src0_ptr += src_stride_3x; src0_ptr += src_stride_3x;
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
@ -1318,7 +1318,7 @@ static void hevc_vt_4t_16w_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src3 = __lsx_vld(src0_ptr, 0); src3 = __lsx_vld(src0_ptr, 0);
src4 = __lsx_vldx(src0_ptr, src_stride); src4 = LSX_VLDX(src0_ptr, src_stride);
src0_ptr += src_stride_2x; src0_ptr += src_stride_2x;
DUP2_ARG2(__lsx_vld, src1_ptr, 0, src1_ptr, 16, in0, in2); DUP2_ARG2(__lsx_vld, src1_ptr, 0, src1_ptr, 16, in0, in2);
src1_ptr += src2_stride; src1_ptr += src2_stride;
@ -1340,7 +1340,7 @@ static void hevc_vt_4t_16w_lsx(uint8_t *src0_ptr, int32_t src_stride,
dst += dst_stride_2x; dst += dst_stride_2x;
src5 = __lsx_vld(src0_ptr, 0); src5 = __lsx_vld(src0_ptr, 0);
src2 = __lsx_vldx(src0_ptr, src_stride); src2 = LSX_VLDX(src0_ptr, src_stride);
src0_ptr += src_stride_2x; src0_ptr += src_stride_2x;
DUP2_ARG2(__lsx_vld, src1_ptr, 0, src1_ptr, 16, in0, in2); DUP2_ARG2(__lsx_vld, src1_ptr, 0, src1_ptr, 16, in0, in2);
src1_ptr += src2_stride; src1_ptr += src2_stride;
@ -1517,7 +1517,7 @@ static void hevc_hv_4t_6w_lsx(uint8_t *src0_ptr, int32_t src_stride,
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src0_ptr += src_stride_3x; src0_ptr += src_stride_3x;
@ -1535,9 +1535,9 @@ static void hevc_hv_4t_6w_lsx(uint8_t *src0_ptr, int32_t src_stride,
DUP2_ARG2(__lsx_vilvh_h, dsth1, dsth0, dsth2, dsth1, tmp1, tmp3); DUP2_ARG2(__lsx_vilvh_h, dsth1, dsth0, dsth2, dsth1, tmp1, tmp3);
src3 = __lsx_vld(src0_ptr, 0); src3 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src0_ptr, src_stride_3x); src6 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, vec0, vec1); DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, vec0, vec1);
DUP2_ARG3(__lsx_vshuf_b, src4, src4, mask0, src4, src4, mask1, vec2, vec3); DUP2_ARG3(__lsx_vshuf_b, src4, src4, mask0, src4, src4, mask1, vec2, vec3);
@ -1550,9 +1550,9 @@ static void hevc_hv_4t_6w_lsx(uint8_t *src0_ptr, int32_t src_stride,
vec5, filt1, dsth6, vec7, filt1, dsth3, dsth4, dsth5, dsth6); vec5, filt1, dsth6, vec7, filt1, dsth3, dsth4, dsth5, dsth6);
src3 = __lsx_vld(src0_ptr, 0); src3 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src0_ptr, src_stride_3x); src6 = LSX_VLDX(src0_ptr, src_stride_3x);
DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, vec0, vec1); DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, vec0, vec1);
DUP2_ARG3(__lsx_vshuf_b, src4, src4, mask0, src4, src4, mask1, vec2, vec3); DUP2_ARG3(__lsx_vshuf_b, src4, src4, mask0, src4, src4, mask1, vec2, vec3);
@ -1700,7 +1700,7 @@ void hevc_hv_4t_8x2_lsx(uint8_t *src0_ptr, int32_t src_stride, int16_t *src1_ptr
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP4_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP4_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src0_ptr, src_stride_3x, src0_ptr, src_stride_4x, src0_ptr, src_stride_3x, src0_ptr, src_stride_4x,
src1, src2, src3, src4); src1, src2, src3, src4);
@ -1777,19 +1777,19 @@ void hevc_hv_4t_8multx4_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (cnt = width8mult; cnt--;) { for (cnt = width8mult; cnt--;) {
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src0_ptr, src_stride_3x); src3 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
src4 = __lsx_vld(src0_ptr, 0); src4 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src5, src6); src5, src6);
src0_ptr += (8 - src_stride_4x); src0_ptr += (8 - src_stride_4x);
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += 8; src1_ptr += 8;
DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1,
@ -1900,22 +1900,22 @@ void hevc_hv_4t_8x6_lsx(uint8_t *src0_ptr, int32_t src_stride, int16_t *src1_ptr
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src0_ptr, 0); src0 = __lsx_vld(src0_ptr, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP2_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src0_ptr, src_stride_3x); src3 = LSX_VLDX(src0_ptr, src_stride_3x);
src0_ptr += src_stride_4x; src0_ptr += src_stride_4x;
src4 = __lsx_vld(src0_ptr, 0); src4 = __lsx_vld(src0_ptr, 0);
DUP4_ARG2(__lsx_vldx, src0_ptr, src_stride, src0_ptr, src_stride_2x, DUP4_ARG2(LSX_VLDX, src0_ptr, src_stride, src0_ptr, src_stride_2x,
src0_ptr, src_stride_3x, src0_ptr, src_stride_4x, src0_ptr, src_stride_3x, src0_ptr, src_stride_4x,
src5, src6, src7, src8); src5, src6, src7, src8);
in0 = __lsx_vld(src1_ptr, 0); in0 = __lsx_vld(src1_ptr, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr, src2_stride_x, src1_ptr, src2_stride_2x, DUP2_ARG2(LSX_VLDX, src1_ptr, src2_stride_x, src1_ptr, src2_stride_2x,
in1, in2); in1, in2);
in3 = __lsx_vldx(src1_ptr, src2_stride_3x); in3 = LSX_VLDX(src1_ptr, src2_stride_3x);
src1_ptr += src2_stride_2x; src1_ptr += src2_stride_2x;
in4 = __lsx_vld(src1_ptr, 0); in4 = __lsx_vld(src1_ptr, 0);
in5 = __lsx_vldx(src1_ptr, src2_stride_x); in5 = LSX_VLDX(src1_ptr, src2_stride_x);
DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, vec0, vec1); DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, vec0, vec1);
DUP2_ARG3(__lsx_vshuf_b, src1, src1, mask0, src1, src1, mask1, vec2, vec3); DUP2_ARG3(__lsx_vshuf_b, src1, src1, mask0, src1, src1, mask1, vec2, vec3);
@ -2041,7 +2041,7 @@ void hevc_hv_4t_8multx4mult_lsx(uint8_t *src0_ptr, int32_t src_stride,
src1_ptr_tmp = src1_ptr; src1_ptr_tmp = src1_ptr;
src0 = __lsx_vld(src0_ptr_tmp, 0); src0 = __lsx_vld(src0_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr_tmp, src_stride, src0_ptr_tmp, DUP2_ARG2(LSX_VLDX, src0_ptr_tmp, src_stride, src0_ptr_tmp,
src_stride_2x, src1, src2); src_stride_2x, src1, src2);
src0_ptr_tmp += src_stride_3x; src0_ptr_tmp += src_stride_3x;
@ -2063,14 +2063,14 @@ void hevc_hv_4t_8multx4mult_lsx(uint8_t *src0_ptr, int32_t src_stride,
for (loop_cnt = height >> 2; loop_cnt--;) { for (loop_cnt = height >> 2; loop_cnt--;) {
src3 = __lsx_vld(src0_ptr_tmp, 0); src3 = __lsx_vld(src0_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src0_ptr_tmp, src_stride, src0_ptr_tmp, DUP2_ARG2(LSX_VLDX, src0_ptr_tmp, src_stride, src0_ptr_tmp,
src_stride_2x, src4, src5); src_stride_2x, src4, src5);
src6 = __lsx_vldx(src0_ptr_tmp, src_stride_3x); src6 = LSX_VLDX(src0_ptr_tmp, src_stride_3x);
src0_ptr_tmp += src_stride_4x; src0_ptr_tmp += src_stride_4x;
in0 = __lsx_vld(src1_ptr_tmp, 0); in0 = __lsx_vld(src1_ptr_tmp, 0);
DUP2_ARG2(__lsx_vldx, src1_ptr_tmp, src2_stride_x, src1_ptr_tmp, DUP2_ARG2(LSX_VLDX, src1_ptr_tmp, src2_stride_x, src1_ptr_tmp,
src2_stride_2x, in1, in2); src2_stride_2x, in1, in2);
in3 = __lsx_vldx(src1_ptr_tmp, src2_stride_3x); in3 = LSX_VLDX(src1_ptr_tmp, src2_stride_3x);
src1_ptr_tmp += src2_stride_2x; src1_ptr_tmp += src2_stride_2x;
DUP4_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, src4, DUP4_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, src4,

76
libavcodec/loongarch/hevc_mc_uni_lsx.c
View File

@ -148,11 +148,11 @@ void common_vt_8t_8w_lsx(uint8_t *src, int32_t src_stride,
filt0, filt1, filt2, filt3); filt0, filt1, filt2, filt3);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += src_stride_3x; src += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
src10_r, src32_r, src54_r, src21_r); src10_r, src32_r, src54_r, src21_r);
@ -160,8 +160,8 @@ void common_vt_8t_8w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10,
@ -228,12 +228,12 @@ void common_vt_8t_16w_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src_tmp, src_stride_3x); src3 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
src4 = __lsx_vld(src_tmp, 0); src4 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src5, src6); src5, src6);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
@ -245,9 +245,9 @@ void common_vt_8t_16w_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src7 = __lsx_vld(src_tmp, 0); src7 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src8, src9); src8, src9);
src10 = __lsx_vldx(src_tmp, src_stride_3x); src10 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10,
src9, src76_r, src87_r, src98_r, src109_r); src9, src76_r, src87_r, src98_r, src109_r);
@ -380,12 +380,12 @@ void hevc_hv_8t_8x2_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src_tmp, src_stride_3x); src3 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
src4 = __lsx_vld(src_tmp, 0); src4 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src5, src6); src5, src6);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -429,7 +429,7 @@ void hevc_hv_8t_8x2_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (loop_cnt = height >> 1; loop_cnt--;) { for (loop_cnt = height >> 1; loop_cnt--;) {
src7 = __lsx_vld(src_tmp, 0); src7 = __lsx_vld(src_tmp, 0);
src8 = __lsx_vldx(src_tmp, src_stride); src8 = LSX_VLDX(src_tmp, src_stride);
src_tmp += src_stride_2x; src_tmp += src_stride_2x;
DUP4_ARG3(__lsx_vshuf_b, src7, src7, mask0, src7, src7, mask1, src7, DUP4_ARG3(__lsx_vshuf_b, src7, src7, mask0, src7, src7, mask1, src7,
@ -567,13 +567,13 @@ void common_vt_4t_24w_lsx(uint8_t *src, int32_t src_stride,
/* 16 width */ /* 16 width */
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l);
/* 8 width */ /* 8 width */
src6 = __lsx_vld(_src, 0); src6 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, src7, src8); DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x, src7, src8);
src += src_stride_3x; src += src_stride_3x;
_src += src_stride_3x; _src += src_stride_3x;
DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r); DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r);
@ -581,7 +581,7 @@ void common_vt_4t_24w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = 8; loop_cnt--;) { for (loop_cnt = 8; loop_cnt--;) {
/* 16 width */ /* 16 width */
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src4, src10); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src4, src10);
DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r); DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r);
DUP2_ARG2(__lsx_vilvh_b, src3, src2, src4, src3, src32_l, src43_l); DUP2_ARG2(__lsx_vilvh_b, src3, src2, src4, src3, src32_l, src43_l);
@ -615,7 +615,7 @@ void common_vt_4t_24w_lsx(uint8_t *src, int32_t src_stride,
/* 16 width */ /* 16 width */
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src5, src11); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src5, src11);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src2, src8); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src2, src8);
DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src5, src4, src2, src5, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src5, src4, src2, src5, src10_l, src21_l);
@ -676,14 +676,14 @@ void common_vt_4t_32w_lsx(uint8_t *src, int32_t src_stride,
/* 16 width */ /* 16 width */
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l);
/* next 16 width */ /* next 16 width */
src6 = __lsx_vld(_src, 0); src6 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, src7, src8); DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x, src7, src8);
src += src_stride_3x; src += src_stride_3x;
_src += src_stride_3x; _src += src_stride_3x;
@ -693,7 +693,7 @@ void common_vt_4t_32w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 1); loop_cnt--;) { for (loop_cnt = (height >> 1); loop_cnt--;) {
/* 16 width */ /* 16 width */
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src4, src10); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src4, src10);
DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r); DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r);
DUP2_ARG2(__lsx_vilvh_b, src3, src2, src4, src3, src32_l, src43_l); DUP2_ARG2(__lsx_vilvh_b, src3, src2, src4, src3, src32_l, src43_l);
@ -774,7 +774,7 @@ void hevc_hv_4t_8x2_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src,
src_stride_3x, src, src_stride_4x, src1, src2, src3, src4); src_stride_3x, src, src_stride_4x, src1, src2, src3, src4);
DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src1, src1, DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src1, src1,
@ -838,11 +838,11 @@ void hevc_hv_4t_8multx4_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (cnt = width8mult; cnt--;) { for (cnt = width8mult; cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += (8 - src_stride_4x); src += (8 - src_stride_4x);
DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1,
vec0, vec1); vec0, vec1);
@ -939,10 +939,10 @@ void hevc_hv_4t_8x6_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x,src, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,src,
src_stride_3x, src, src_stride_4x, src1, src2, src3, src4); src_stride_3x, src, src_stride_4x, src1, src2, src3, src4);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x,src, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,src,
src_stride_3x, src, src_stride_4x, src5, src6, src7, src8); src_stride_3x, src, src_stride_4x, src5, src6, src7, src8);
DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src1, src1, DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src1, src1,
@ -1051,7 +1051,7 @@ void hevc_hv_4t_8multx4mult_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -1073,9 +1073,9 @@ void hevc_hv_4t_8multx4mult_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src3 = __lsx_vld(src_tmp, 0); src3 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src_tmp, src_stride_3x); src6 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, src4, DUP4_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, src4,
@ -1185,7 +1185,7 @@ void hevc_hv_4t_12w_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -1204,9 +1204,9 @@ void hevc_hv_4t_12w_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (loop_cnt = 4; loop_cnt--;) { for (loop_cnt = 4; loop_cnt--;) {
src3 = __lsx_vld(src_tmp, 0); src3 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src_tmp, src_stride_3x); src6 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, src4, DUP4_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, src4,
@ -1260,7 +1260,7 @@ void hevc_hv_4t_12w_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
mask3 = __lsx_vaddi_bu(mask2, 2); mask3 = __lsx_vaddi_bu(mask2, 2);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src += src_stride_3x; src += src_stride_3x;
DUP2_ARG3(__lsx_vshuf_b, src1, src0, mask2, src1, src0, mask3, vec0, vec1); DUP2_ARG3(__lsx_vshuf_b, src1, src0, mask2, src1, src0, mask3, vec0, vec1);
DUP2_ARG3(__lsx_vshuf_b, src2, src1, mask2, src2, src1, mask3, vec2, vec3); DUP2_ARG3(__lsx_vshuf_b, src2, src1, mask2, src2, src1, mask3, vec2, vec3);
@ -1275,12 +1275,12 @@ void hevc_hv_4t_12w_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (loop_cnt = 2; loop_cnt--;) { for (loop_cnt = 2; loop_cnt--;) {
src3 = __lsx_vld(src, 0); src3 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src4, src5); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src4, src5);
src6 = __lsx_vldx(src, src_stride_3x); src6 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src7, src3, mask2, src7, src3, mask3, src8, DUP4_ARG3(__lsx_vshuf_b, src7, src3, mask2, src7, src3, mask3, src8,
src4, mask2, src8, src4, mask3, vec0, vec1, vec2, vec3); src4, mask2, src8, src4, mask3, vec0, vec1, vec2, vec3);

8
libavcodec/loongarch/hevc_mc_uniw_lsx.c
View File

@ -79,12 +79,12 @@ void hevc_hv_8t_8x2_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src_tmp, src_stride_3x); src3 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
src4 = __lsx_vld(src_tmp, 0); src4 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src5, src6); src5, src6);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -127,7 +127,7 @@ void hevc_hv_8t_8x2_lsx(uint8_t *src, int32_t src_stride, uint8_t *dst,
for (loop_cnt = height >> 1; loop_cnt--;) { for (loop_cnt = height >> 1; loop_cnt--;) {
src7 = __lsx_vld(src_tmp, 0); src7 = __lsx_vld(src_tmp, 0);
src8 = __lsx_vldx(src_tmp, src_stride); src8 = LSX_VLDX(src_tmp, src_stride);
src_tmp += src_stride_2x; src_tmp += src_stride_2x;
DUP4_ARG3(__lsx_vshuf_b, src7, src7, mask0, src7, src7, mask1, src7, DUP4_ARG3(__lsx_vshuf_b, src7, src7, mask0, src7, src7, mask1, src7,
src7, mask2, src7, src7, mask3, vec0, vec1, vec2, vec3); src7, mask2, src7, src7, mask3, vec0, vec1, vec2, vec3);

218
libavcodec/loongarch/hevcdsp_lsx.c
View File

@ -48,14 +48,14 @@ static void hevc_copy_4w_lsx(uint8_t *src, int32_t src_stride,
__m128i in0, in1, in2, in3; __m128i in0, in1, in2, in3;
for (; loop_cnt--;) { for (; loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src5, src6); src5, src6);
src7 = __lsx_vldx(src, src_stride_3x); src7 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_w, src1, src0, src3, src2, src5, src4, src7, src6, DUP4_ARG2(__lsx_vilvl_w, src1, src0, src3, src2, src5, src4, src7, src6,
@ -98,12 +98,12 @@ static void hevc_copy_6w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 3); loop_cnt--;) { for (loop_cnt = (height >> 3); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src7 = __lsx_vldx(src, src_stride_3x); src7 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6, DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6,
@ -163,14 +163,14 @@ static void hevc_copy_8w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 3); loop_cnt--;) { for (loop_cnt = (height >> 3); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src5, src6); src5, src6);
src7 = __lsx_vldx(src, src_stride_3x); src7 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6, DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6,
@ -215,12 +215,12 @@ static void hevc_copy_12w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 3); loop_cnt--;) { for (loop_cnt = (height >> 3); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src7 = __lsx_vldx(src, src_stride_3x); src7 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6, DUP4_ARG2(__lsx_vsllwil_hu_bu, src0, 6, src1, 6, src2, 6, src3, 6,
@ -288,14 +288,14 @@ static void hevc_copy_16w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 3); loop_cnt--;) { for (loop_cnt = (height >> 3); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src5, src6); src5, src6);
src7 = __lsx_vldx(src, src_stride_3x); src7 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero, src3, DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero, src3,
in0_l, in1_l, in2_l, in3_l); in0_l, in1_l, in2_l, in3_l);
@ -333,8 +333,8 @@ static void hevc_copy_16w_lsx(uint8_t *src, int32_t src_stride,
} }
if (res) { if (res) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero, src3, DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero, src3,
in0_l, in1_l, in2_l, in3_l); in0_l, in1_l, in2_l, in3_l);
@ -373,13 +373,13 @@ static void hevc_copy_24w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(_src, 0); src4 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x,
src5, src6); src5, src6);
src7 = __lsx_vldx(_src, src_stride_3x); src7 = LSX_VLDX(_src, src_stride_3x);
_src += src_stride_4x; _src += src_stride_4x;
DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero, DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero,
@ -423,13 +423,13 @@ static void hevc_copy_32w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src2, src4); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src2, src4);
src6 = __lsx_vldx(src, src_stride_3x); src6 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src1 = __lsx_vld(_src, 0); src1 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x,
src3, src5); src3, src5);
src7 = __lsx_vldx(_src, src_stride_3x); src7 = LSX_VLDX(_src, src_stride_3x);
_src += src_stride_4x; _src += src_stride_4x;
DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero, DUP4_ARG2(__lsx_vilvh_b, zero, src0, zero, src1, zero, src2, zero,
@ -623,12 +623,12 @@ static void hevc_hz_8t_4w_lsx(uint8_t *src, int32_t src_stride,
for (;loop_cnt--;) { for (;loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src7 = __lsx_vldx(src, src_stride_3x); src7 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src1, src0, mask0, src1, src0, mask1, src1, DUP4_ARG3(__lsx_vshuf_b, src1, src0, mask0, src1, src0, mask1, src1,
src0, mask2, src1, src0, mask3, vec0, vec1, vec2, vec3); src0, mask2, src1, src0, mask3, vec0, vec1, vec2, vec3);
@ -668,7 +668,7 @@ static void hevc_hz_8t_4w_lsx(uint8_t *src, int32_t src_stride,
} }
for (;res--;) { for (;res--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
src1 = __lsx_vldx(src, src_stride); src1 = LSX_VLDX(src, src_stride);
DUP4_ARG3(__lsx_vshuf_b, src1, src0, mask0, src1, src0, mask1, src1, DUP4_ARG3(__lsx_vshuf_b, src1, src0, mask0, src1, src0, mask1, src1,
src0, mask2, src1, src0, mask3, vec0, vec1, vec2, vec3); src0, mask2, src1, src0, mask3, vec0, vec1, vec2, vec3);
dst0 = __lsx_vdp2_h_bu_b(vec0, filt0); dst0 = __lsx_vdp2_h_bu_b(vec0, filt0);
@ -709,8 +709,8 @@ static void hevc_hz_8t_8w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src0, DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src0,
@ -774,12 +774,12 @@ static void hevc_hz_8t_12w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = 4; loop_cnt--;) { for (loop_cnt = 4; loop_cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src4 = __lsx_vld(_src, 0); src4 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x,
src5, src6); src5, src6);
src7 = __lsx_vldx(_src, src_stride_3x); src7 = LSX_VLDX(_src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
_src += src_stride_4x; _src += src_stride_4x;
@ -1216,11 +1216,11 @@ static void hevc_vt_8t_4w_lsx(uint8_t *src, int32_t src_stride,
filt0, filt1, filt2, filt3); filt0, filt1, filt2, filt3);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += src_stride_3x; src += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
src10_r, src32_r, src54_r, src21_r); src10_r, src32_r, src54_r, src21_r);
@ -1231,13 +1231,13 @@ static void hevc_vt_8t_4w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 3); loop_cnt--;) { for (loop_cnt = (height >> 3); loop_cnt--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src11 = __lsx_vld(src, 0); src11 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x,
src12, src13); src12, src13);
src14 = __lsx_vldx(src, src_stride_3x); src14 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, src9, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, src9,
@ -1289,7 +1289,7 @@ static void hevc_vt_8t_4w_lsx(uint8_t *src, int32_t src_stride,
} }
for (;res--;) { for (;res--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
src8 = __lsx_vldx(src, src_stride); src8 = LSX_VLDX(src, src_stride);
DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r); DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r);
src += src_stride_2x; src += src_stride_2x;
src8776 = __lsx_vilvl_d(src87_r, src76_r); src8776 = __lsx_vilvl_d(src87_r, src76_r);
@ -1334,11 +1334,11 @@ static void hevc_vt_8t_8w_lsx(uint8_t *src, int32_t src_stride,
filt0, filt1, filt2, filt3); filt0, filt1, filt2, filt3);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += src_stride_3x; src += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
src10_r, src32_r, src54_r, src21_r); src10_r, src32_r, src54_r, src21_r);
@ -1346,8 +1346,8 @@ static void hevc_vt_8t_8w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10,
src9, src76_r, src87_r, src98_r, src109_r); src9, src76_r, src87_r, src98_r, src109_r);
@ -1408,11 +1408,11 @@ static void hevc_vt_8t_12w_lsx(uint8_t *src, int32_t src_stride,
DUP4_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filter, 4, filter, 6, DUP4_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filter, 4, filter, 6,
filt0, filt1, filt2, filt3); filt0, filt1, filt2, filt3);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += src_stride_3x; src += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
src10_r, src32_r, src54_r, src21_r); src10_r, src32_r, src54_r, src21_r);
@ -1426,8 +1426,8 @@ static void hevc_vt_8t_12w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, src10,
src9, src76_r, src87_r, src98_r, src109_r); src9, src76_r, src87_r, src98_r, src109_r);
@ -1520,12 +1520,12 @@ static void hevc_vt_8t_16multx4mult_lsx(uint8_t *src,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src_tmp, src_stride_3x); src3 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
src4 = __lsx_vld(src_tmp, 0); src4 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src5, src6); src5, src6);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1, DUP4_ARG2(__lsx_vilvl_b, src1, src0, src3, src2, src5, src4, src2, src1,
@ -1537,9 +1537,9 @@ static void hevc_vt_8t_16multx4mult_lsx(uint8_t *src,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src7 = __lsx_vld(src_tmp, 0); src7 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src8, src9); src8, src9);
src10 = __lsx_vldx(src_tmp, src_stride_3x); src10 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8, DUP4_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src9, src8,
src10, src9, src76_r, src87_r, src98_r, src109_r); src10, src9, src76_r, src87_r, src98_r, src109_r);
@ -1689,11 +1689,11 @@ static void hevc_hv_8t_4w_lsx(uint8_t *src, int32_t src_stride,
mask3 = __lsx_vaddi_bu(mask0, 6); mask3 = __lsx_vaddi_bu(mask0, 6);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += src_stride_3x; src += src_stride_3x;
DUP4_ARG3(__lsx_vshuf_b, src3, src0, mask0, src3, src0, mask1, src3, src0, DUP4_ARG3(__lsx_vshuf_b, src3, src0, mask0, src3, src0, mask1, src3, src0,
@ -1729,8 +1729,8 @@ static void hevc_hv_8t_4w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = height >> 2; loop_cnt--;) { for (loop_cnt = height >> 2; loop_cnt--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src9, src7, mask0, src9, src7, mask1, src9, src7, DUP4_ARG3(__lsx_vshuf_b, src9, src7, mask0, src9, src7, mask1, src9, src7,
@ -1830,12 +1830,12 @@ static void hevc_hv_8t_8multx1mult_lsx(uint8_t *src,
src_tmp = src; src_tmp = src;
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src_tmp, src_stride_3x); src3 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
src4 = __lsx_vld(src_tmp, 0); src4 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src5, src6); src5, src6);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -1978,12 +1978,12 @@ static void hevc_hv_8t_12w_lsx(uint8_t *src, int32_t src_stride,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src3 = __lsx_vldx(src_tmp, src_stride_3x); src3 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
src4 = __lsx_vld(src_tmp, 0); src4 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src5, src6); src5, src6);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -2077,11 +2077,11 @@ static void hevc_hv_8t_12w_lsx(uint8_t *src, int32_t src_stride,
mask7 = __lsx_vaddi_bu(mask4, 6); mask7 = __lsx_vaddi_bu(mask4, 6);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src4 = __lsx_vld(src, 0); src4 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += src_stride_3x; src += src_stride_3x;
DUP4_ARG3(__lsx_vshuf_b, src3, src0, mask4, src3, src0, mask5, src3, src0, DUP4_ARG3(__lsx_vshuf_b, src3, src0, mask4, src3, src0, mask5, src3, src0,
@ -2118,8 +2118,8 @@ static void hevc_hv_8t_12w_lsx(uint8_t *src, int32_t src_stride,
for (loop_cnt = height >> 2; loop_cnt--;) { for (loop_cnt = height >> 2; loop_cnt--;) {
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG3(__lsx_vshuf_b, src9, src7, mask4, src9, src7, mask5, src9, DUP4_ARG3(__lsx_vshuf_b, src9, src7, mask4, src9, src7, mask5, src9,
@ -2285,14 +2285,14 @@ static void hevc_vt_4t_16w_lsx(uint8_t *src,
DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1); DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src += src_stride_3x; src += src_stride_3x;
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l);
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
src3 = __lsx_vld(src, 0); src3 = __lsx_vld(src, 0);
src4 = __lsx_vldx(src, src_stride); src4 = LSX_VLDX(src, src_stride);
src += src_stride_2x; src += src_stride_2x;
DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r); DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r);
DUP2_ARG2(__lsx_vilvh_b, src3, src2, src4, src3, src32_l, src43_l); DUP2_ARG2(__lsx_vilvh_b, src3, src2, src4, src3, src32_l, src43_l);
@ -2309,7 +2309,7 @@ static void hevc_vt_4t_16w_lsx(uint8_t *src,
dst += dst_stride; dst += dst_stride;
src5 = __lsx_vld(src, 0); src5 = __lsx_vld(src, 0);
src2 = __lsx_vldx(src, src_stride); src2 = LSX_VLDX(src, src_stride);
src += src_stride_2x; src += src_stride_2x;
DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src5, src4, src2, src5, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src5, src4, src2, src5, src10_l, src21_l);
@ -2353,19 +2353,19 @@ static void hevc_vt_4t_24w_lsx(uint8_t *src,
DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1); DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l);
src6 = __lsx_vld(_src, 0); src6 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, src7, src8); DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x, src7, src8);
src += src_stride_3x; src += src_stride_3x;
_src += src_stride_3x; _src += src_stride_3x;
DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r); DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r);
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src4, src10); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src4, src10);
src += src_stride_2x; src += src_stride_2x;
_src += src_stride_2x; _src += src_stride_2x;
DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r); DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r);
@ -2392,7 +2392,7 @@ static void hevc_vt_4t_24w_lsx(uint8_t *src,
dst += dst_stride; dst += dst_stride;
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src5, src11); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src5, src11);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src2, src8); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src2, src8);
src += src_stride_2x; src += src_stride_2x;
_src += src_stride_2x; _src += src_stride_2x;
DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r);
@ -2448,12 +2448,12 @@ static void hevc_vt_4t_32w_lsx(uint8_t *src,
DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1); DUP2_ARG2(__lsx_vldrepl_h, filter, 0, filter, 2, filt0, filt1);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src1, src0, src2, src1, src10_r, src21_r);
DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l); DUP2_ARG2(__lsx_vilvh_b, src1, src0, src2, src1, src10_l, src21_l);
src6 = __lsx_vld(_src, 0); src6 = __lsx_vld(_src, 0);
DUP2_ARG2(__lsx_vldx, _src, src_stride, _src, src_stride_2x, src7, src8); DUP2_ARG2(LSX_VLDX, _src, src_stride, _src, src_stride_2x, src7, src8);
src += src_stride_3x; src += src_stride_3x;
_src += src_stride_3x; _src += src_stride_3x;
DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r); DUP2_ARG2(__lsx_vilvl_b, src7, src6, src8, src7, src76_r, src87_r);
@ -2461,7 +2461,7 @@ static void hevc_vt_4t_32w_lsx(uint8_t *src,
for (loop_cnt = (height >> 2); loop_cnt--;) { for (loop_cnt = (height >> 2); loop_cnt--;) {
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src3, src9);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src4, src10); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src4, src10);
src += src_stride_2x; src += src_stride_2x;
_src += src_stride_2x; _src += src_stride_2x;
DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r); DUP2_ARG2(__lsx_vilvl_b, src3, src2, src4, src3, src32_r, src43_r);
@ -2493,7 +2493,7 @@ static void hevc_vt_4t_32w_lsx(uint8_t *src,
dst += dst_stride; dst += dst_stride;
DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src5, src11); DUP2_ARG2(__lsx_vld, src, 0, _src, 0, src5, src11);
DUP2_ARG2(__lsx_vldx, src, src_stride, _src, src_stride, src2, src8); DUP2_ARG2(LSX_VLDX, src, src_stride, _src, src_stride, src2, src8);
src += src_stride_2x; src += src_stride_2x;
_src += src_stride_2x; _src += src_stride_2x;
DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r); DUP2_ARG2(__lsx_vilvl_b, src5, src4, src2, src5, src10_r, src21_r);
@ -2560,9 +2560,9 @@ static void hevc_hv_4t_8x2_lsx(uint8_t *src,
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src3 = __lsx_vldx(src, src_stride_3x); src3 = LSX_VLDX(src, src_stride_3x);
src4 = __lsx_vldx(src, src_stride_4x); src4 = LSX_VLDX(src, src_stride_4x);
DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, vec0, vec1); DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, vec0, vec1);
DUP2_ARG3(__lsx_vshuf_b, src1, src1, mask0, src1, src1, mask1, vec2, vec3); DUP2_ARG3(__lsx_vshuf_b, src1, src1, mask0, src1, src1, mask1, vec2, vec3);
@ -2627,10 +2627,10 @@ static void hevc_hv_4t_8multx4_lsx(uint8_t *src, int32_t src_stride,
for (cnt = width8mult; cnt--;) { for (cnt = width8mult; cnt--;) {
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src,
src_stride_3x, src, src_stride_4x, src1, src2, src3, src4); src_stride_3x, src, src_stride_4x, src1, src2, src3, src4);
src += src_stride_4x; src += src_stride_4x;
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src5, src6); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src5, src6);
src += (8 - src_stride_4x); src += (8 - src_stride_4x);
DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, DUP2_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1,
@ -2730,10 +2730,10 @@ static void hevc_hv_4t_8x6_lsx(uint8_t *src,
mask1 = __lsx_vaddi_bu(mask0, 2); mask1 = __lsx_vaddi_bu(mask0, 2);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src,
src_stride_3x, src, src_stride_4x, src1, src2, src3, src4); src_stride_3x, src, src_stride_4x, src1, src2, src3, src4);
src += src_stride_4x; src += src_stride_4x;
DUP4_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src, DUP4_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src,
src_stride_3x, src, src_stride_4x, src5, src6, src7, src8); src_stride_3x, src, src_stride_4x, src5, src6, src7, src8);
DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src1, src1, DUP4_ARG3(__lsx_vshuf_b, src0, src0, mask0, src0, src0, mask1, src1, src1,
@ -2847,7 +2847,7 @@ static void hevc_hv_4t_8multx4mult_lsx(uint8_t *src,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -2869,9 +2869,9 @@ static void hevc_hv_4t_8multx4mult_lsx(uint8_t *src,
for (loop_cnt = height >> 2; loop_cnt--;) { for (loop_cnt = height >> 2; loop_cnt--;) {
src3 = __lsx_vld(src_tmp, 0); src3 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src_tmp, src_stride_3x); src6 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1,
@ -2997,7 +2997,7 @@ static void hevc_hv_4t_12w_lsx(uint8_t *src,
dst_tmp = dst; dst_tmp = dst;
src0 = __lsx_vld(src_tmp, 0); src0 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src1, src2); src1, src2);
src_tmp += src_stride_3x; src_tmp += src_stride_3x;
@ -3016,9 +3016,9 @@ static void hevc_hv_4t_12w_lsx(uint8_t *src,
for (loop_cnt = 4; loop_cnt--;) { for (loop_cnt = 4; loop_cnt--;) {
src3 = __lsx_vld(src_tmp, 0); src3 = __lsx_vld(src_tmp, 0);
DUP2_ARG2(__lsx_vldx, src_tmp, src_stride, src_tmp, src_stride_2x, DUP2_ARG2(LSX_VLDX, src_tmp, src_stride, src_tmp, src_stride_2x,
src4, src5); src4, src5);
src6 = __lsx_vldx(src_tmp, src_stride_3x); src6 = LSX_VLDX(src_tmp, src_stride_3x);
src_tmp += src_stride_4x; src_tmp += src_stride_4x;
DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1, DUP2_ARG3(__lsx_vshuf_b, src3, src3, mask0, src3, src3, mask1,
@ -3077,7 +3077,7 @@ static void hevc_hv_4t_12w_lsx(uint8_t *src,
mask3 = __lsx_vaddi_bu(mask2, 2); mask3 = __lsx_vaddi_bu(mask2, 2);
src0 = __lsx_vld(src, 0); src0 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src1, src2); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src1, src2);
src += src_stride_3x; src += src_stride_3x;
DUP2_ARG3(__lsx_vshuf_b, src1, src0, mask2, src1, src0, mask3, vec0, vec1); DUP2_ARG3(__lsx_vshuf_b, src1, src0, mask2, src1, src0, mask3, vec0, vec1);
DUP2_ARG3(__lsx_vshuf_b, src2, src1, mask2, src2, src1, mask3, vec2, vec3); DUP2_ARG3(__lsx_vshuf_b, src2, src1, mask2, src2, src1, mask3, vec2, vec3);
@ -3090,12 +3090,12 @@ static void hevc_hv_4t_12w_lsx(uint8_t *src,
for (loop_cnt = 2; loop_cnt--;) { for (loop_cnt = 2; loop_cnt--;) {
src3 = __lsx_vld(src, 0); src3 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src4, src5); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src4, src5);
src6 = __lsx_vldx(src, src_stride_3x); src6 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
src7 = __lsx_vld(src, 0); src7 = __lsx_vld(src, 0);
DUP2_ARG2(__lsx_vldx, src, src_stride, src, src_stride_2x, src8, src9); DUP2_ARG2(LSX_VLDX, src, src_stride, src, src_stride_2x, src8, src9);
src10 = __lsx_vldx(src, src_stride_3x); src10 = LSX_VLDX(src, src_stride_3x);
src += src_stride_4x; src += src_stride_4x;
DUP2_ARG3(__lsx_vshuf_b, src7, src3, mask2, src7, src3, mask3, DUP2_ARG3(__lsx_vshuf_b, src7, src3, mask2, src7, src3, mask3,
vec0, vec1); vec0, vec1);

5
libavutil/loongarch/loongson_intrinsics.h
View File

@ -89,6 +89,11 @@
#ifdef __loongarch_sx #ifdef __loongarch_sx
#include <lsxintrin.h> #include <lsxintrin.h>
/* __lsx_vldx() from lsxintrin.h does not accept a const void*;
* remove the following once it does. */
#define LSX_VLDX(cptr, stride) __lsx_vldx((void*)(cptr), (stride))
/* /*
* ============================================================================= * =============================================================================
* Description : Dot product & addition of byte vector elements * Description : Dot product & addition of byte vector elements