mirror of https://git.ffmpeg.org/ffmpeg.git
avoid alignment hacks, luckly gcc does the right thing on arches different from x86
Originally committed as revision 5915 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
parent
3b0df5254e
commit
8047fe72a9
|
@ -71,8 +71,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc00_ ## CODETYPE (uint8_t *dst, uin
|
|||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc10_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){ \
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/8]);\
|
||||
uint8_t * const half= (uint8_t*)temp;\
|
||||
DECLARE_ALIGNED_16(uint8_t, half[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src, half, stride, stride, SIZE);\
|
||||
}\
|
||||
|
@ -82,15 +81,13 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc20_ ## CODETYPE(uint8_t *dst, uint
|
|||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc30_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/8]);\
|
||||
uint8_t * const half= (uint8_t*)temp;\
|
||||
DECLARE_ALIGNED_16(uint8_t, half[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src+1, half, stride, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc01_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/8]);\
|
||||
uint8_t * const half= (uint8_t*)temp;\
|
||||
DECLARE_ALIGNED_16(uint8_t, half[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src, half, stride, stride, SIZE);\
|
||||
}\
|
||||
|
@ -100,89 +97,79 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc02_ ## CODETYPE(uint8_t *dst, uint
|
|||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc03_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/8]);\
|
||||
uint8_t * const half= (uint8_t*)temp;\
|
||||
DECLARE_ALIGNED_16(uint8_t, half[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src+stride, half, stride, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc11_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/4]);\
|
||||
uint8_t * const halfH= (uint8_t*)temp;\
|
||||
uint8_t * const halfV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfH[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfV[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc31_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/4]);\
|
||||
uint8_t * const halfH= (uint8_t*)temp;\
|
||||
uint8_t * const halfV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfH[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfV[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc13_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/4]);\
|
||||
uint8_t * const halfH= (uint8_t*)temp;\
|
||||
uint8_t * const halfV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfH[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfV[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc33_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*SIZE/4]);\
|
||||
uint8_t * const halfH= (uint8_t*)temp;\
|
||||
uint8_t * const halfV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfH[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfV[SIZE*SIZE]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc22_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*(SIZE+8)/4]);\
|
||||
int16_t * const tmp= (int16_t*)temp;\
|
||||
DECLARE_ALIGNED_16(int16_t, tmp[SIZE*(SIZE+8)]);\
|
||||
OPNAME ## h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(dst, tmp, src, stride, SIZE, stride);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc21_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*(SIZE+8)/4 + SIZE*SIZE/4]);\
|
||||
uint8_t * const halfH= (uint8_t*)temp;\
|
||||
uint8_t * const halfHV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
int16_t * const tmp= ((int16_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfH[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfHV[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(int16_t, tmp[SIZE*(SIZE+8)]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc23_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*(SIZE+8)/4 + SIZE*SIZE/4]);\
|
||||
uint8_t * const halfH= (uint8_t*)temp;\
|
||||
uint8_t * const halfHV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
int16_t * const tmp= ((int16_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfH[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfHV[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(int16_t, tmp[SIZE*(SIZE+8)]);\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc12_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*(SIZE+8)/4 + SIZE*SIZE/4]);\
|
||||
uint8_t * const halfV= (uint8_t*)temp;\
|
||||
uint8_t * const halfHV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
int16_t * const tmp= ((int16_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfV[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfHV[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(int16_t, tmp[SIZE*(SIZE+8)]);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfV, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc32_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED_16(uint64_t, temp[SIZE*(SIZE+8)/4 + SIZE*SIZE/4]);\
|
||||
uint8_t * const halfV= (uint8_t*)temp;\
|
||||
uint8_t * const halfHV= ((uint8_t*)temp) + SIZE*SIZE;\
|
||||
int16_t * const tmp= ((int16_t*)temp) + SIZE*SIZE;\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfV[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(uint8_t, halfHV[SIZE*SIZE]);\
|
||||
DECLARE_ALIGNED_16(int16_t, tmp[SIZE*(SIZE+8)]);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfV, halfHV, stride, SIZE, SIZE);\
|
||||
|
|
Loading…
Reference in New Issue