@@ -24,7 +24,7 @@
#include <stddef.h>
#include <stdint.h>
-void ff_hevc_sao_band_filter_8x8_8_neon(uint8_t *_dst, const uint8_t *_src,
+void ff_h26x_sao_band_filter_8x8_8_neon(uint8_t *_dst, const uint8_t *_src,
ptrdiff_t stride_dst, ptrdiff_t stride_src,
const int16_t *sao_offset_val, int sao_left_class,
int width, int height);
@@ -33,4 +33,8 @@ void ff_hevc_sao_edge_filter_16x16_8_neon(uint8_t *dst, const uint8_t *src, ptrd
void ff_hevc_sao_edge_filter_8x8_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst,
const int16_t *sao_offset_val, int eo, int width, int height);
+void ff_vvc_sao_edge_filter_16x16_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst,
+ const int16_t *sao_offset_val, int eo, int width, int height);
+void ff_vvc_sao_edge_filter_8x8_8_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride_dst,
+ const int16_t *sao_offset_val, int eo, int width, int height);
#endif
@@ -24,15 +24,17 @@
#include "libavutil/aarch64/asm.S"
-#define MAX_PB_SIZE 64
+#define HEVC_MAX_PB_SIZE 64
+#define VVC_MAX_PB_SIZE 128
#define AV_INPUT_BUFFER_PADDING_SIZE 64
-#define SAO_STRIDE (2*MAX_PB_SIZE + AV_INPUT_BUFFER_PADDING_SIZE)
+#define HEVC_SAO_STRIDE (2 * HEVC_MAX_PB_SIZE + AV_INPUT_BUFFER_PADDING_SIZE)
+#define VVC_SAO_STRIDE (2 * VVC_MAX_PB_SIZE + AV_INPUT_BUFFER_PADDING_SIZE)
// void sao_band_filter(uint8_t *_dst, uint8_t *_src,
// ptrdiff_t stride_dst, ptrdiff_t stride_src,
// int16_t *sao_offset_val, int sao_left_class,
// int width, int height)
-function ff_hevc_sao_band_filter_8x8_8_neon, export=1
+function ff_h26x_sao_band_filter_8x8_8_neon, export=1
stp xzr, xzr, [sp, #-64]!
stp xzr, xzr, [sp, #16]
stp xzr, xzr, [sp, #32]
@@ -79,16 +81,30 @@ function ff_hevc_sao_band_filter_8x8_8_neon, export=1
ret
endfunc
-.Lsao_edge_pos:
+.Lhevc_sao_edge_pos:
.word 1 // horizontal
-.word SAO_STRIDE // vertical
-.word SAO_STRIDE + 1 // 45 degree
-.word SAO_STRIDE - 1 // 135 degree
+.word HEVC_SAO_STRIDE // vertical
+.word HEVC_SAO_STRIDE + 1 // 45 degree
+.word HEVC_SAO_STRIDE - 1 // 135 degree
+
+.Lvvc_sao_edge_pos:
+.word 1 // horizontal
+.word VVC_SAO_STRIDE // vertical
+.word VVC_SAO_STRIDE + 1 // 45 degree
+.word VVC_SAO_STRIDE - 1 // 135 degree
+
+function ff_vvc_sao_edge_filter_16x16_8_neon, export=1
+ adr x7, .Lvvc_sao_edge_pos
+ mov x15, #VVC_SAO_STRIDE
+ b 1f
+endfunc
// ff_hevc_sao_edge_filter_16x16_8_neon(char *dst, char *src, ptrdiff stride_dst,
// int16 *sao_offset_val, int eo, int width, int height)
function ff_hevc_sao_edge_filter_16x16_8_neon, export=1
- adr x7, .Lsao_edge_pos
+ adr x7, .Lhevc_sao_edge_pos
+ mov x15, #HEVC_SAO_STRIDE
+1:
ld1 {v3.8h}, [x3] // load sao_offset_val
add w5, w5, #0xF
bic w5, w5, #0xF
@@ -101,7 +117,6 @@ function ff_hevc_sao_edge_filter_16x16_8_neon, export=1
uzp2 v1.16b, v3.16b, v3.16b // sao_offset_val -> upper
uzp1 v0.16b, v3.16b, v3.16b // sao_offset_val -> lower
movi v2.16b, #2
- mov x15, #SAO_STRIDE
// strides between end of line and next src/dst
sub x15, x15, x5 // stride_src - width
sub x16, x2, x5 // stride_dst - width
@@ -145,10 +160,18 @@ function ff_hevc_sao_edge_filter_16x16_8_neon, export=1
ret
endfunc
+function ff_vvc_sao_edge_filter_8x8_8_neon, export=1
+ adr x7, .Lvvc_sao_edge_pos
+ mov x15, #VVC_SAO_STRIDE
+ b 1f
+endfunc
+
// ff_hevc_sao_edge_filter_8x8_8_neon(char *dst, char *src, ptrdiff stride_dst,
// int16 *sao_offset_val, int eo, int width, int height)
function ff_hevc_sao_edge_filter_8x8_8_neon, export=1
- adr x7, .Lsao_edge_pos
+ adr x7, .Lhevc_sao_edge_pos
+ mov x15, #HEVC_SAO_STRIDE
+1:
ldr w4, [x7, w4, uxtw #2]
ld1 {v3.8h}, [x3]
mov v3.h[7], v3.h[0]
@@ -160,7 +183,6 @@ function ff_hevc_sao_edge_filter_8x8_8_neon, export=1
movi v2.16b, #2
add x16, x0, x2
lsl x2, x2, #1
- mov x15, #SAO_STRIDE
mov x8, x1
sub x9, x1, x4
add x10, x1, x4
@@ -384,7 +384,7 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
c->sao_band_filter[1] =
c->sao_band_filter[2] =
c->sao_band_filter[3] =
- c->sao_band_filter[4] = ff_hevc_sao_band_filter_8x8_8_neon;
+ c->sao_band_filter[4] = ff_h26x_sao_band_filter_8x8_8_neon;
c->sao_edge_filter[0] = ff_hevc_sao_edge_filter_8x8_8_neon;
c->sao_edge_filter[1] =
c->sao_edge_filter[2] =
@@ -1,5 +1,6 @@
clean::
$(RM) $(CLEANSUFFIXES:%=libavcodec/aarch64/vvc/%)
-OBJS-$(CONFIG_VVC_DECODER) += aarch64/vvc/dsp_init.o
-NEON-OBJS-$(CONFIG_VVC_DECODER) += aarch64/vvc/alf.o
+OBJS-$(CONFIG_VVC_DECODER) += aarch64/vvc/dsp_init.o
+NEON-OBJS-$(CONFIG_VVC_DECODER) += aarch64/vvc/alf.o \
+ aarch64/h26x/sao_neon.o
@@ -22,6 +22,7 @@
#include "libavutil/cpu.h"
#include "libavutil/aarch64/cpu.h"
+#include "libavcodec/aarch64/h26x/dsp.h"
#include "libavcodec/vvc/dsp.h"
#include "libavcodec/vvc/dec.h"
#include "libavcodec/vvc/ctu.h"
@@ -45,6 +46,11 @@ void ff_vvc_dsp_init_aarch64(VVCDSPContext *const c, const int bd)
return;
if (bd == 8) {
+ for (int i = 0; i < FF_ARRAY_ELEMS(c->sao.band_filter); i++)
+ c->sao.band_filter[i] = ff_h26x_sao_band_filter_8x8_8_neon;
+ c->sao.edge_filter[0] = ff_vvc_sao_edge_filter_8x8_8_neon;
+ for (int i = 1; i < FF_ARRAY_ELEMS(c->sao.edge_filter); i++)
+ c->sao.edge_filter[i] = ff_vvc_sao_edge_filter_16x16_8_neon;
c->alf.filter[LUMA] = alf_filter_luma_8_neon;
c->alf.filter[CHROMA] = alf_filter_chroma_8_neon;
} else if (bd == 10) {
From: Zhao Zhili <zhilizhao@tencent.com> --- libavcodec/aarch64/h26x/dsp.h | 6 +++- libavcodec/aarch64/h26x/sao_neon.S | 44 +++++++++++++++++------ libavcodec/aarch64/hevcdsp_init_aarch64.c | 2 +- libavcodec/aarch64/vvc/Makefile | 5 +-- libavcodec/aarch64/vvc/dsp_init.c | 6 ++++ 5 files changed, 48 insertions(+), 15 deletions(-)