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

Optimize pure C unscaled yuv2rgb.

Browse Source 
Aligning the tables reduces the amount of code generated on
e.g. ARM as the offset constant then has few enough set bits
so it can be encoded inside a single instruction instead of 2.
Ideally all should be declared aligned, but the DECLARE_ALIGNED
macros does not work with pointer tables, thus also reordered
the tables.

Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
This commit is contained in:
Reimar Döffinger 2013-10-25 22:46:06 +02:00
parent 18802942d1
commit 4fab08c94f
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
libswscale/swscale_internal.h
View File

@ -362,9 +362,11 @@ typedef struct SwsContext {
int dstY; ///< Last destination vertical line output from last slice. int dstY; ///< Last destination vertical line output from last slice.
int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc... int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
void *yuvTable; // pointer to the yuv->rgb table start so it can be freed() void *yuvTable; // pointer to the yuv->rgb table start so it can be freed()
// alignment ensures the offset can be added in a single
// instruction on e.g. ARM
DECLARE_ALIGNED(16, int, table_gV)[256 + 2*YUVRGB_TABLE_HEADROOM];
uint8_t *table_rV[256 + 2*YUVRGB_TABLE_HEADROOM]; uint8_t *table_rV[256 + 2*YUVRGB_TABLE_HEADROOM];
uint8_t *table_gU[256 + 2*YUVRGB_TABLE_HEADROOM]; uint8_t *table_gU[256 + 2*YUVRGB_TABLE_HEADROOM];
int table_gV[256 + 2*YUVRGB_TABLE_HEADROOM];
uint8_t *table_bU[256 + 2*YUVRGB_TABLE_HEADROOM]; uint8_t *table_bU[256 + 2*YUVRGB_TABLE_HEADROOM];
DECLARE_ALIGNED(16, int32_t, input_rgb2yuv_table)[16+40*4]; // This table can contain both C and SIMD formatted values, teh C vales are always at the XY_IDX points DECLARE_ALIGNED(16, int32_t, input_rgb2yuv_table)[16+40*4]; // This table can contain both C and SIMD formatted values, teh C vales are always at the XY_IDX points
#define RY_IDX 0 #define RY_IDX 0