| Message ID | 20240609092709.1356010-1-remi@remlab.net |
|---|---|
| State | New |
| Headers | show |
| Series | [FFmpeg-devel,PATCHv2,1/4] lavc/h263dsp: add DCT dequantisation function | expand |
| Context | Check | Description |
|---|---|---|
| andriy/make_x86 | success | Make finished |
| andriy/make_fate_x86 | success | Make fate finished |
Le sunnuntaina 9. kesäkuuta 2024, 12.27.06 EEST Rémi Denis-Courmont a écrit : > To preserve the alignment promise of the block base address, we pass > the start offset separately (it is either 0 or 1). > > Note that optimised implementations of this function will be taken into > actual use if MpegEncContext.dct_unquantize_h263_{inter,intra} are *not* > overloaded by existing optimisations. > --- > libavcodec/h263dsp.c | 17 +++++++++++++++++ > libavcodec/h263dsp.h | 2 ++ > 2 files changed, 19 insertions(+) > > diff --git a/libavcodec/h263dsp.c b/libavcodec/h263dsp.c > index 6a13353499..eb990f27bd 100644 > --- a/libavcodec/h263dsp.c > +++ b/libavcodec/h263dsp.c > @@ -23,6 +23,22 @@ > #include "config.h" > #include "h263dsp.h" > > +static void h263_dct_unquantize_c(int16_t *block, size_t start, size_t end, > + int qmul, int qadd) > +{ > + for (size_t i = start; i <= end; i++) { C and R-V V don't care, but for the sake of more constrained SIMD extensions, it should be easier to keep intra and inter DCT functions separate.
diff --git a/libavcodec/h263dsp.c b/libavcodec/h263dsp.c index 6a13353499..eb990f27bd 100644 --- a/libavcodec/h263dsp.c +++ b/libavcodec/h263dsp.c @@ -23,6 +23,22 @@ #include "config.h" #include "h263dsp.h" +static void h263_dct_unquantize_c(int16_t *block, size_t start, size_t end, + int qmul, int qadd) +{ + for (size_t i = start; i <= end; i++) { + int level = block[i]; + + if (level) { + if (level < 0) + level = level * qmul - qadd; + else + level = level * qmul + qadd; + block[i] = level; + } + } +} + const uint8_t ff_h263_loop_filter_strength[32] = { 0, 1, 1, 2, 2, 3, 3, 4, 4, 4, 5, 5, 6, 6, 7, 7, 7, 8, 8, 8, 9, 9, 9, 10, 10, 10, 11, 11, 11, 12, 12, 12 @@ -116,6 +132,7 @@ static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale) av_cold void ff_h263dsp_init(H263DSPContext *ctx) { + ctx->h263_dct_unquantize = h263_dct_unquantize_c; ctx->h263_h_loop_filter = h263_h_loop_filter_c; ctx->h263_v_loop_filter = h263_v_loop_filter_c; diff --git a/libavcodec/h263dsp.h b/libavcodec/h263dsp.h index 2dccd23392..93c128c9ac 100644 --- a/libavcodec/h263dsp.h +++ b/libavcodec/h263dsp.h @@ -24,6 +24,8 @@ extern const uint8_t ff_h263_loop_filter_strength[32]; typedef struct H263DSPContext { + void (*h263_dct_unquantize)(int16_t *block /* align 16 */, size_t offset, + size_t len, int mul, int add); void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale); void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale); } H263DSPContext;
To preserve the alignment promise of the block base address, we pass the start offset separately (it is either 0 or 1). Note that optimised implementations of this function will be taken into actual use if MpegEncContext.dct_unquantize_h263_{inter,intra} are *not* overloaded by existing optimisations. --- libavcodec/h263dsp.c | 17 +++++++++++++++++ libavcodec/h263dsp.h | 2 ++ 2 files changed, 19 insertions(+)