merge two ITransforms together when applicable and change the TTransform

to return the sum directly.

output is bitwise the same, speed up 1-2%. This is preparatory to a
more efficient SSE2 implementation.

Change-Id: I0bcdf05808c93420fbe9dcb75e5e7e55a4ae5b89
This commit is contained in:
Pascal Massimino
2011-04-21 13:32:45 -07:00
parent ca554137d2
commit e7ff3f9af6
3 changed files with 35 additions and 24 deletions

View File

@ -416,7 +416,10 @@ int VP8Decimate(VP8EncIterator* const it, VP8ModeScore* const rd, int rd_opt);
// in dsp.c
// Transforms
typedef void (*VP8Idct)(const uint8_t* ref, const int16_t* in, uint8_t* dst);
// VP8Idct: Does one of two inverse transforms. If do_two is set, the transforms
// will be done for (ref, in, dst) and (ref + 4, in + 16, dst + 4).
typedef void (*VP8Idct)(const uint8_t* ref, const int16_t* in, uint8_t* dst,
int do_two);
typedef void (*VP8Fdct)(const uint8_t* src, const uint8_t* ref, int16_t* out);
typedef void (*VP8WHT)(const int16_t* in, int16_t* out);
extern VP8Idct VP8ITransform;