har0ke/FFmpeg
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

avcodec/mips: Improve hevc sao band filter msa functions

Browse Source 
Preload data in band filter 0-8 for better pipeline parallelization.

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit is contained in:
Kaustubh Raste 2017-09-15 17:06:34 +05:30 committed by Michael Niedermayer
parent 0105ed551c
commit 1a85fb7e1e
2 changed files with 112 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

174
libavcodec/mips/hevc_lpf_sao_msa.c
View File

@ -1049,29 +1049,28 @@ static void hevc_sao_band_filter_4width_msa(uint8_t *dst, int32_t dst_stride,
int16_t *sao_offset_val, int16_t *sao_offset_val,
int32_t height) int32_t height)
{ {
int32_t h_cnt;
v16u8 src0, src1, src2, src3; v16u8 src0, src1, src2, src3;
v16i8 src0_r, src1_r; v16i8 src0_r, src1_r;
v16i8 offset, offset_val, mask; v16i8 offset, offset_val, mask;
v16i8 offset0 = { 0 }; v16i8 dst0, offset0, offset1;
v16i8 offset1 = { 0 };
v16i8 zero = { 0 }; v16i8 zero = { 0 };
v8i16 temp0, temp1, dst0, dst1;
offset_val = LD_SB(sao_offset_val + 1); offset_val = LD_SB(sao_offset_val + 1);
offset_val = (v16i8) __msa_pckev_d((v2i64) offset_val, (v2i64) offset_val); offset_val = (v16i8) __msa_pckev_d((v2i64) offset_val, (v2i64) offset_val);
offset_val = __msa_pckev_b(offset_val, offset_val); offset_val = __msa_pckev_b(offset_val, offset_val);
offset1 = (v16i8) __msa_insve_w((v4i32) offset1, 3, (v4i32) offset_val); offset1 = (v16i8) __msa_insve_w((v4i32) zero, 3, (v4i32) offset_val);
offset0 = __msa_sld_b(offset1, offset0, 28 - ((sao_left_class) & 31)); offset0 = __msa_sld_b(offset1, zero, 28 - ((sao_left_class) & 31));
offset1 = __msa_sld_b(zero, offset1, 28 - ((sao_left_class) & 31)); offset1 = __msa_sld_b(zero, offset1, 28 - ((sao_left_class) & 31));
/* load in advance. */
LD_UB4(src, src_stride, src0, src1, src2, src3);
if (!((sao_left_class > 12) & (sao_left_class < 29))) { if (!((sao_left_class > 12) & (sao_left_class < 29))) {
SWAP(offset0, offset1); SWAP(offset0, offset1);
} }
for (h_cnt = height >> 2; h_cnt--;) { for (height -= 4; height; height -= 4) {
LD_UB4(src, src_stride, src0, src1, src2, src3);
src += (4 * src_stride); src += (4 * src_stride);
ILVEV_D2_SB(src0, src1, src2, src3, src0_r, src1_r); ILVEV_D2_SB(src0, src1, src2, src3, src0_r, src1_r);
@ -1080,14 +1079,30 @@ static void hevc_sao_band_filter_4width_msa(uint8_t *dst, int32_t dst_stride,
mask = __msa_srli_b(src0_r, 3); mask = __msa_srli_b(src0_r, 3);
offset = __msa_vshf_b(mask, offset1, offset0); offset = __msa_vshf_b(mask, offset1, offset0);
UNPCK_SB_SH(offset, temp0, temp1); src0_r = (v16i8) __msa_xori_b((v16u8) src0_r, 128);
ILVRL_B2_SH(zero, src0_r, dst0, dst1); dst0 = __msa_adds_s_b(src0_r, offset);
ADD2(dst0, temp0, dst1, temp1, dst0, dst1); dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128);
CLIP_SH2_0_255(dst0, dst1);
dst0 = (v8i16) __msa_pckev_b((v16i8) dst1, (v16i8) dst0); /* load in advance. */
LD_UB4(src, src_stride, src0, src1, src2, src3);
/* store results */
ST4x4_UB(dst0, dst0, 0, 1, 2, 3, dst, dst_stride); ST4x4_UB(dst0, dst0, 0, 1, 2, 3, dst, dst_stride);
dst += (4 * dst_stride); dst += (4 * dst_stride);
} }
ILVEV_D2_SB(src0, src1, src2, src3, src0_r, src1_r);
src0_r = (v16i8) __msa_pckev_w((v4i32) src1_r, (v4i32) src0_r);
mask = __msa_srli_b(src0_r, 3);
offset = __msa_vshf_b(mask, offset1, offset0);
src0_r = (v16i8) __msa_xori_b((v16u8) src0_r, 128);
dst0 = __msa_adds_s_b(src0_r, offset);
dst0 = (v16i8) __msa_xori_b((v16u8) dst0, 128);
/* store results */
ST4x4_UB(dst0, dst0, 0, 1, 2, 3, dst, dst_stride);
} }
static void hevc_sao_band_filter_8width_msa(uint8_t *dst, int32_t dst_stride, static void hevc_sao_band_filter_8width_msa(uint8_t *dst, int32_t dst_stride,
@ -1096,51 +1111,69 @@ static void hevc_sao_band_filter_8width_msa(uint8_t *dst, int32_t dst_stride,
int16_t *sao_offset_val, int16_t *sao_offset_val,
int32_t height) int32_t height)
{ {
int32_t h_cnt;
v16u8 src0, src1, src2, src3; v16u8 src0, src1, src2, src3;
v16i8 src0_r, src1_r, mask0, mask1; v16i8 src0_r, src1_r, mask0, mask1;
v16i8 offset, offset_val; v16i8 offset_mask0, offset_mask1, offset_val;
v16i8 offset0 = { 0 }; v16i8 offset0, offset1, dst0, dst1;
v16i8 offset1 = { 0 };
v16i8 zero = { 0 }; v16i8 zero = { 0 };
v8i16 dst0, dst1, dst2, dst3;
v8i16 temp0, temp1, temp2, temp3;
offset_val = LD_SB(sao_offset_val + 1); offset_val = LD_SB(sao_offset_val + 1);
offset_val = (v16i8) __msa_pckev_d((v2i64) offset_val, (v2i64) offset_val); offset_val = (v16i8) __msa_pckev_d((v2i64) offset_val, (v2i64) offset_val);
offset_val = __msa_pckev_b(offset_val, offset_val); offset_val = __msa_pckev_b(offset_val, offset_val);
offset1 = (v16i8) __msa_insve_w((v4i32) offset1, 3, (v4i32) offset_val); offset1 = (v16i8) __msa_insve_w((v4i32) zero, 3, (v4i32) offset_val);
offset0 = __msa_sld_b(offset1, offset0, 28 - ((sao_left_class) & 31)); offset0 = __msa_sld_b(offset1, zero, 28 - ((sao_left_class) & 31));
offset1 = __msa_sld_b(zero, offset1, 28 - ((sao_left_class) & 31)); offset1 = __msa_sld_b(zero, offset1, 28 - ((sao_left_class) & 31));
/* load in advance. */
LD_UB4(src, src_stride, src0, src1, src2, src3);
if (!((sao_left_class > 12) & (sao_left_class < 29))) { if (!((sao_left_class > 12) & (sao_left_class < 29))) {
SWAP(offset0, offset1); SWAP(offset0, offset1);
} }
for (h_cnt = height >> 2; h_cnt--;) { for (height -= 4; height; height -= 4) {
LD_UB4(src, src_stride, src0, src1, src2, src3); src += src_stride << 2;
src += (4 * src_stride);
ILVR_D2_SB(src1, src0, src3, src2, src0_r, src1_r); ILVR_D2_SB(src1, src0, src3, src2, src0_r, src1_r);
mask0 = __msa_srli_b(src0_r, 3); mask0 = __msa_srli_b(src0_r, 3);
mask1 = __msa_srli_b(src1_r, 3); mask1 = __msa_srli_b(src1_r, 3);
offset = __msa_vshf_b(mask0, offset1, offset0); offset_mask0 = __msa_vshf_b(mask0, offset1, offset0);
UNPCK_SB_SH(offset, temp0, temp1); offset_mask1 = __msa_vshf_b(mask1, offset1, offset0);
offset = __msa_vshf_b(mask1, offset1, offset0); /* load in advance. */
UNPCK_SB_SH(offset, temp2, temp3); LD_UB4(src, src_stride, src0, src1, src2, src3);
UNPCK_UB_SH(src0_r, dst0, dst1); XORI_B2_128_SB(src0_r, src1_r);
UNPCK_UB_SH(src1_r, dst2, dst3);
ADD4(dst0, temp0, dst1, temp1, dst2, temp2, dst3, temp3, dst0 = __msa_adds_s_b(src0_r, offset_mask0);
dst0, dst1, dst2, dst3); dst1 = __msa_adds_s_b(src1_r, offset_mask1);
CLIP_SH4_0_255(dst0, dst1, dst2, dst3);
PCKEV_B2_SH(dst1, dst0, dst3, dst2, dst0, dst2); XORI_B2_128_SB(dst0, dst1);
ST8x4_UB(dst0, dst2, dst, dst_stride);
dst += (4 * dst_stride); /* store results */
ST8x4_UB(dst0, dst1, dst, dst_stride);
dst += dst_stride << 2;
} }
ILVR_D2_SB(src1, src0, src3, src2, src0_r, src1_r);
mask0 = __msa_srli_b(src0_r, 3);
mask1 = __msa_srli_b(src1_r, 3);
offset_mask0 = __msa_vshf_b(mask0, offset1, offset0);
offset_mask1 = __msa_vshf_b(mask1, offset1, offset0);
XORI_B2_128_SB(src0_r, src1_r);
dst0 = __msa_adds_s_b(src0_r, offset_mask0);
dst1 = __msa_adds_s_b(src1_r, offset_mask1);
XORI_B2_128_SB(dst0, dst1);
/* store results */
ST8x4_UB(dst0, dst1, dst, dst_stride);
} }
static void hevc_sao_band_filter_16multiple_msa(uint8_t *dst, static void hevc_sao_band_filter_16multiple_msa(uint8_t *dst,
@ -1151,32 +1184,30 @@ static void hevc_sao_band_filter_16multiple_msa(uint8_t *dst,
int16_t *sao_offset_val, int16_t *sao_offset_val,
int32_t width, int32_t height) int32_t width, int32_t height)
{ {
int32_t h_cnt, w_cnt; int32_t w_cnt;
v16u8 src0, src1, src2, src3; v16u8 src0, src1, src2, src3;
v8i16 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;
v16i8 out0, out1, out2, out3; v16i8 out0, out1, out2, out3;
v16i8 mask0, mask1, mask2, mask3; v16i8 mask0, mask1, mask2, mask3;
v16i8 tmp0, tmp1, tmp2, tmp3, offset_val; v16i8 tmp0, tmp1, tmp2, tmp3, offset_val;
v16i8 offset0 = { 0 }; v16i8 offset0, offset1;
v16i8 offset1 = { 0 };
v16i8 zero = { 0 }; v16i8 zero = { 0 };
v8i16 temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
offset_val = LD_SB(sao_offset_val + 1); offset_val = LD_SB(sao_offset_val + 1);
offset_val = (v16i8) __msa_pckev_d((v2i64) offset_val, (v2i64) offset_val); offset_val = (v16i8) __msa_pckev_d((v2i64) offset_val, (v2i64) offset_val);
offset_val = __msa_pckev_b(offset_val, offset_val); offset_val = __msa_pckev_b(offset_val, offset_val);
offset1 = (v16i8) __msa_insve_w((v4i32) offset1, 3, (v4i32) offset_val); offset1 = (v16i8) __msa_insve_w((v4i32) zero, 3, (v4i32) offset_val);
offset0 = __msa_sld_b(offset1, offset0, 28 - ((sao_left_class) & 31)); offset0 = __msa_sld_b(offset1, zero, 28 - ((sao_left_class) & 31));
offset1 = __msa_sld_b(zero, offset1, 28 - ((sao_left_class) & 31)); offset1 = __msa_sld_b(zero, offset1, 28 - ((sao_left_class) & 31));
if (!((sao_left_class > 12) & (sao_left_class < 29))) { if (!((sao_left_class > 12) & (sao_left_class < 29))) {
SWAP(offset0, offset1); SWAP(offset0, offset1);
} }
for (h_cnt = height >> 2; h_cnt--;) { while (height > 0) {
for (w_cnt = 0; w_cnt < (width >> 4); w_cnt++) { /* load in advance */
LD_UB4(src + w_cnt * 16, src_stride, src0, src1, src2, src3); LD_UB4(src, src_stride, src0, src1, src2, src3);
for (w_cnt = 16; w_cnt < width; w_cnt += 16) {
mask0 = __msa_srli_b((v16i8) src0, 3); mask0 = __msa_srli_b((v16i8) src0, 3);
mask1 = __msa_srli_b((v16i8) src1, 3); mask1 = __msa_srli_b((v16i8) src1, 3);
mask2 = __msa_srli_b((v16i8) src2, 3); mask2 = __msa_srli_b((v16i8) src2, 3);
@ -1186,27 +1217,44 @@ static void hevc_sao_band_filter_16multiple_msa(uint8_t *dst,
tmp0, tmp1); tmp0, tmp1);
VSHF_B2_SB(offset0, offset1, offset0, offset1, mask2, mask3, VSHF_B2_SB(offset0, offset1, offset0, offset1, mask2, mask3,
tmp2, tmp3); tmp2, tmp3);
UNPCK_SB_SH(tmp0, temp0, temp1); XORI_B4_128_UB(src0, src1, src2, src3);
UNPCK_SB_SH(tmp1, temp2, temp3);
UNPCK_SB_SH(tmp2, temp4, temp5); out0 = __msa_adds_s_b((v16i8) src0, tmp0);
UNPCK_SB_SH(tmp3, temp6, temp7); out1 = __msa_adds_s_b((v16i8) src1, tmp1);
ILVRL_B2_SH(zero, src0, dst0, dst1); out2 = __msa_adds_s_b((v16i8) src2, tmp2);
ILVRL_B2_SH(zero, src1, dst2, dst3); out3 = __msa_adds_s_b((v16i8) src3, tmp3);
ILVRL_B2_SH(zero, src2, dst4, dst5);
ILVRL_B2_SH(zero, src3, dst6, dst7); /* load for next iteration */
ADD4(dst0, temp0, dst1, temp1, dst2, temp2, dst3, temp3, LD_UB4(src + w_cnt, src_stride, src0, src1, src2, src3);
dst0, dst1, dst2, dst3);
ADD4(dst4, temp4, dst5, temp5, dst6, temp6, dst7, temp7, XORI_B4_128_SB(out0, out1, out2, out3);
dst4, dst5, dst6, dst7);
CLIP_SH4_0_255(dst0, dst1, dst2, dst3); ST_SB4(out0, out1, out2, out3, dst + w_cnt - 16, dst_stride);
CLIP_SH4_0_255(dst4, dst5, dst6, dst7);
PCKEV_B4_SB(dst1, dst0, dst3, dst2, dst5, dst4, dst7, dst6,
out0, out1, out2, out3);
ST_SB4(out0, out1, out2, out3, dst + w_cnt * 16, dst_stride);
} }
mask0 = __msa_srli_b((v16i8) src0, 3);
mask1 = __msa_srli_b((v16i8) src1, 3);
mask2 = __msa_srli_b((v16i8) src2, 3);
mask3 = __msa_srli_b((v16i8) src3, 3);
VSHF_B2_SB(offset0, offset1, offset0, offset1, mask0, mask1, tmp0,
tmp1);
VSHF_B2_SB(offset0, offset1, offset0, offset1, mask2, mask3, tmp2,
tmp3);
XORI_B4_128_UB(src0, src1, src2, src3);
out0 = __msa_adds_s_b((v16i8) src0, tmp0);
out1 = __msa_adds_s_b((v16i8) src1, tmp1);
out2 = __msa_adds_s_b((v16i8) src2, tmp2);
out3 = __msa_adds_s_b((v16i8) src3, tmp3);
XORI_B4_128_SB(out0, out1, out2, out3);
ST_SB4(out0, out1, out2, out3, dst + w_cnt - 16, dst_stride);
src += src_stride << 2; src += src_stride << 2;
dst += dst_stride << 2; dst += dst_stride << 2;
height -= 4;
} }
} }

1
libavutil/mips/generic_macros_msa.h
View File

@ -1574,6 +1574,7 @@
out0 = (RTYPE) __msa_ilvr_h((v8i16) in0, (v8i16) in1); \ out0 = (RTYPE) __msa_ilvr_h((v8i16) in0, (v8i16) in1); \
out1 = (RTYPE) __msa_ilvl_h((v8i16) in0, (v8i16) in1); \ out1 = (RTYPE) __msa_ilvl_h((v8i16) in0, (v8i16) in1); \
} }
#define ILVRL_H2_UB(...) ILVRL_H2(v16u8, __VA_ARGS__)
#define ILVRL_H2_SB(...) ILVRL_H2(v16i8, __VA_ARGS__) #define ILVRL_H2_SB(...) ILVRL_H2(v16i8, __VA_ARGS__)
#define ILVRL_H2_SH(...) ILVRL_H2(v8i16, __VA_ARGS__) #define ILVRL_H2_SH(...) ILVRL_H2(v8i16, __VA_ARGS__)
#define ILVRL_H2_SW(...) ILVRL_H2(v4i32, __VA_ARGS__) #define ILVRL_H2_SW(...) ILVRL_H2(v4i32, __VA_ARGS__)