Commit 297e2768 authored by Derek Buitenhuis's avatar Derek Buitenhuis

Merge commit 'ad9d3384'

* commit 'ad9d3384':
  svq3: move the dequant buffer to SVQ3Context
Merged-by: 's avatarDerek Buitenhuis <derek.buitenhuis@gmail.com>
parents 80d14de5 ad9d3384
...@@ -43,23 +43,6 @@ ...@@ -43,23 +43,6 @@
#include "rectangle.h" #include "rectangle.h"
#include "thread.h" #include "thread.h"
static const uint8_t rem6[QP_MAX_NUM + 1] = {
0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
0, 1, 2, 3,
};
static const uint8_t div6[QP_MAX_NUM + 1] = {
0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3,
3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6,
7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10,
10,10,10,11,11,11,11,11,11,12,12,12,12,12,12,13,13,13, 13, 13, 13,
14,14,14,14,
};
static const uint8_t field_scan[16+1] = { static const uint8_t field_scan[16+1] = {
0 + 0 * 4, 0 + 1 * 4, 1 + 0 * 4, 0 + 2 * 4, 0 + 0 * 4, 0 + 1 * 4, 1 + 0 * 4, 0 + 2 * 4,
0 + 3 * 4, 1 + 1 * 4, 1 + 2 * 4, 1 + 3 * 4, 0 + 3 * 4, 1 + 1 * 4, 1 + 2 * 4, 1 + 3 * 4,
...@@ -125,29 +108,6 @@ static const uint8_t zigzag_scan8x8_cavlc[64+1] = { ...@@ -125,29 +108,6 @@ static const uint8_t zigzag_scan8x8_cavlc[64+1] = {
5 + 5 * 8, 6 + 5 * 8, 6 + 6 * 8, 7 + 7 * 8, 5 + 5 * 8, 6 + 5 * 8, 6 + 6 * 8, 7 + 7 * 8,
}; };
static const uint8_t dequant4_coeff_init[6][3] = {
{ 10, 13, 16 },
{ 11, 14, 18 },
{ 13, 16, 20 },
{ 14, 18, 23 },
{ 16, 20, 25 },
{ 18, 23, 29 },
};
static const uint8_t dequant8_coeff_init_scan[16] = {
0, 3, 4, 3, 3, 1, 5, 1, 4, 5, 2, 5, 3, 1, 5, 1
};
static const uint8_t dequant8_coeff_init[6][6] = {
{ 20, 18, 32, 19, 25, 24 },
{ 22, 19, 35, 21, 28, 26 },
{ 26, 23, 42, 24, 33, 31 },
{ 28, 25, 45, 26, 35, 33 },
{ 32, 28, 51, 30, 40, 38 },
{ 36, 32, 58, 34, 46, 43 },
};
static void release_unused_pictures(H264Context *h, int remove_current) static void release_unused_pictures(H264Context *h, int remove_current)
{ {
int i; int i;
...@@ -328,11 +288,11 @@ static void init_dequant8_coeff_table(H264Context *h) ...@@ -328,11 +288,11 @@ static void init_dequant8_coeff_table(H264Context *h)
continue; continue;
for (q = 0; q < max_qp + 1; q++) { for (q = 0; q < max_qp + 1; q++) {
int shift = div6[q]; int shift = ff_h264_quant_div6[q];
int idx = rem6[q]; int idx = ff_h264_quant_rem6[q];
for (x = 0; x < 64; x++) for (x = 0; x < 64; x++)
h->dequant8_coeff[i][q][(x >> 3) | ((x & 7) << 3)] = h->dequant8_coeff[i][q][(x >> 3) | ((x & 7) << 3)] =
((uint32_t)dequant8_coeff_init[idx][dequant8_coeff_init_scan[((x >> 1) & 12) | (x & 3)]] * ((uint32_t)ff_h264_dequant8_coeff_init[idx][ff_h264_dequant8_coeff_init_scan[((x >> 1) & 12) | (x & 3)]] *
h->pps.scaling_matrix8[i][x]) << shift; h->pps.scaling_matrix8[i][x]) << shift;
} }
} }
...@@ -354,11 +314,11 @@ static void init_dequant4_coeff_table(H264Context *h) ...@@ -354,11 +314,11 @@ static void init_dequant4_coeff_table(H264Context *h)
continue; continue;
for (q = 0; q < max_qp + 1; q++) { for (q = 0; q < max_qp + 1; q++) {
int shift = div6[q] + 2; int shift = ff_h264_quant_div6[q] + 2;
int idx = rem6[q]; int idx = ff_h264_quant_rem6[q];
for (x = 0; x < 16; x++) for (x = 0; x < 16; x++)
h->dequant4_coeff[i][q][(x >> 2) | ((x << 2) & 0xF)] = h->dequant4_coeff[i][q][(x >> 2) | ((x << 2) & 0xF)] =
((uint32_t)dequant4_coeff_init[idx][(x & 1) + ((x >> 2) & 1)] * ((uint32_t)ff_h264_dequant4_coeff_init[idx][(x & 1) + ((x >> 2) & 1)] *
h->pps.scaling_matrix4[i][x]) << shift; h->pps.scaling_matrix4[i][x]) << shift;
} }
} }
......
...@@ -147,3 +147,41 @@ const PMbInfo ff_h264_b_sub_mb_type_info[13] = { ...@@ -147,3 +147,41 @@ const PMbInfo ff_h264_b_sub_mb_type_info[13] = {
{ MB_TYPE_8x8 | MB_TYPE_P0L1 | MB_TYPE_P1L1, 4, }, { MB_TYPE_8x8 | MB_TYPE_P0L1 | MB_TYPE_P1L1, 4, },
{ MB_TYPE_8x8 | MB_TYPE_P0L0 | MB_TYPE_P0L1 | MB_TYPE_P1L0 | MB_TYPE_P1L1, 4, }, { MB_TYPE_8x8 | MB_TYPE_P0L0 | MB_TYPE_P0L1 | MB_TYPE_P1L0 | MB_TYPE_P1L1, 4, },
}; };
const uint8_t ff_h264_dequant4_coeff_init[6][3] = {
{ 10, 13, 16 },
{ 11, 14, 18 },
{ 13, 16, 20 },
{ 14, 18, 23 },
{ 16, 20, 25 },
{ 18, 23, 29 },
};
const uint8_t ff_h264_dequant8_coeff_init_scan[16] = {
0, 3, 4, 3, 3, 1, 5, 1, 4, 5, 2, 5, 3, 1, 5, 1
};
const uint8_t ff_h264_dequant8_coeff_init[6][6] = {
{ 20, 18, 32, 19, 25, 24 },
{ 22, 19, 35, 21, 28, 26 },
{ 26, 23, 42, 24, 33, 31 },
{ 28, 25, 45, 26, 35, 33 },
{ 32, 28, 51, 30, 40, 38 },
{ 36, 32, 58, 34, 46, 43 },
};
const uint8_t ff_h264_quant_rem6[QP_MAX_NUM + 1] = {
0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
0, 1, 2, 3,
};
const uint8_t ff_h264_quant_div6[QP_MAX_NUM + 1] = {
0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3,
3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6,
7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10,
10,10,10,11,11,11,11,11,11,12,12,12,12,12,12,13,13,13, 13, 13, 13,
14,14,14,14,
};
...@@ -67,4 +67,11 @@ static const AVRational ff_h264_pixel_aspect[17] = { ...@@ -67,4 +67,11 @@ static const AVRational ff_h264_pixel_aspect[17] = {
{ 3, 2 }, { 3, 2 },
{ 2, 1 }, { 2, 1 },
}; };
extern const uint8_t ff_h264_dequant4_coeff_init[6][3];
extern const uint8_t ff_h264_dequant8_coeff_init_scan[16];
extern const uint8_t ff_h264_dequant8_coeff_init[6][6];
extern const uint8_t ff_h264_quant_rem6[QP_MAX_NUM + 1];
extern const uint8_t ff_h264_quant_div6[QP_MAX_NUM + 1];
#endif /* AVCODEC_H264DATA_H */ #endif /* AVCODEC_H264DATA_H */
...@@ -113,6 +113,7 @@ typedef struct SVQ3Context { ...@@ -113,6 +113,7 @@ typedef struct SVQ3Context {
DECLARE_ALIGNED(16, int16_t, mv_cache)[2][5 * 8][2]; DECLARE_ALIGNED(16, int16_t, mv_cache)[2][5 * 8][2];
DECLARE_ALIGNED(8, int8_t, ref_cache)[2][5 * 8]; DECLARE_ALIGNED(8, int8_t, ref_cache)[2][5 * 8];
uint32_t dequant4_coeff[QP_MAX_NUM + 1][16];
} SVQ3Context; } SVQ3Context;
#define FULLPEL_MODE 1 #define FULLPEL_MODE 1
...@@ -663,8 +664,6 @@ static void hl_decode_mb(SVQ3Context *s, const H264Context *h, H264SliceContext ...@@ -663,8 +664,6 @@ static void hl_decode_mb(SVQ3Context *s, const H264Context *h, H264SliceContext
s->vdsp.prefetch(dest_y + (s->mb_x & 3) * 4 * sl->linesize + 64, sl->linesize, 4); s->vdsp.prefetch(dest_y + (s->mb_x & 3) * 4 * sl->linesize + 64, sl->linesize, 4);
s->vdsp.prefetch(dest_cb + (s->mb_x & 7) * sl->uvlinesize + 64, dest_cr - dest_cb, 2); s->vdsp.prefetch(dest_cb + (s->mb_x & 7) * sl->uvlinesize + 64, dest_cr - dest_cb, 2);
h->list_counts[mb_xy] = sl->list_count;
linesize = sl->mb_linesize = sl->linesize; linesize = sl->mb_linesize = sl->linesize;
uvlinesize = sl->mb_uvlinesize = sl->uvlinesize; uvlinesize = sl->mb_uvlinesize = sl->uvlinesize;
...@@ -680,9 +679,9 @@ static void hl_decode_mb(SVQ3Context *s, const H264Context *h, H264SliceContext ...@@ -680,9 +679,9 @@ static void hl_decode_mb(SVQ3Context *s, const H264Context *h, H264SliceContext
if (sl->cbp & 0x30) { if (sl->cbp & 0x30) {
uint8_t *dest[2] = { dest_cb, dest_cr }; uint8_t *dest[2] = { dest_cb, dest_cr };
s->h264dsp.h264_chroma_dc_dequant_idct(sl->mb + 16 * 16 * 1, s->h264dsp.h264_chroma_dc_dequant_idct(sl->mb + 16 * 16 * 1,
h->dequant4_coeff[IS_INTRA(mb_type) ? 1 : 4][sl->chroma_qp[0]][0]); s->dequant4_coeff[sl->chroma_qp[0]][0]);
s->h264dsp.h264_chroma_dc_dequant_idct(sl->mb + 16 * 16 * 2, s->h264dsp.h264_chroma_dc_dequant_idct(sl->mb + 16 * 16 * 2,
h->dequant4_coeff[IS_INTRA(mb_type) ? 2 : 5][sl->chroma_qp[1]][0]); s->dequant4_coeff[sl->chroma_qp[1]][0]);
for (j = 1; j < 3; j++) { for (j = 1; j < 3; j++) {
for (i = j * 16; i < j * 16 + 4; i++) for (i = j * 16; i < j * 16 + 4; i++)
if (sl->non_zero_count_cache[scan8[i]] || sl->mb[i * 16]) { if (sl->non_zero_count_cache[scan8[i]] || sl->mb[i * 16]) {
...@@ -1093,6 +1092,20 @@ static int svq3_decode_slice_header(AVCodecContext *avctx) ...@@ -1093,6 +1092,20 @@ static int svq3_decode_slice_header(AVCodecContext *avctx)
return 0; return 0;
} }
static void init_dequant4_coeff_table(SVQ3Context *s)
{
int q, x;
const int max_qp = 51;
for (q = 0; q < max_qp + 1; q++) {
int shift = ff_h264_quant_div6[q] + 2;
int idx = ff_h264_quant_rem6[q];
for (x = 0; x < 16; x++)
s->dequant4_coeff[q][(x >> 2) | ((x << 2) & 0xF)] =
((uint32_t)ff_h264_dequant4_coeff_init[idx][(x & 1) + ((x >> 2) & 1)] * 16) << shift;
}
}
static av_cold int svq3_decode_init(AVCodecContext *avctx) static av_cold int svq3_decode_init(AVCodecContext *avctx)
{ {
SVQ3Context *s = avctx->priv_data; SVQ3Context *s = avctx->priv_data;
...@@ -1132,8 +1145,6 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx) ...@@ -1132,8 +1145,6 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx)
ff_h264_pred_init(&s->hpc, AV_CODEC_ID_SVQ3, 8, 1); ff_h264_pred_init(&s->hpc, AV_CODEC_ID_SVQ3, 8, 1);
ff_videodsp_init(&s->vdsp, 8); ff_videodsp_init(&s->vdsp, 8);
memset(h->pps.scaling_matrix4, 16, 6 * 16 * sizeof(uint8_t));
memset(h->pps.scaling_matrix8, 16, 2 * 64 * sizeof(uint8_t));
avctx->bits_per_raw_sample = 8; avctx->bits_per_raw_sample = 8;
h->sps.bit_depth_luma = 8; h->sps.bit_depth_luma = 8;
...@@ -1322,10 +1333,7 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx) ...@@ -1322,10 +1333,7 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx)
s->mb2br_xy[mb_xy] = 8 * (mb_xy % (2 * h->mb_stride)); s->mb2br_xy[mb_xy] = 8 * (mb_xy % (2 * h->mb_stride));
} }
if ((ret = ff_h264_alloc_tables(h)) < 0) { init_dequant4_coeff_table(s);
av_log(avctx, AV_LOG_ERROR, "svq3 memory allocation failed\n");
goto fail;
}
return 0; return 0;
fail: fail:
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment