virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
Code Issues Releases Activity

Merge remote-tracking branch 'qatar/master'

Browse Source 
* qatar/master:
  ppc: reduce overreads when loading 8 pixels in altivec dsp functions

Merged-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Michael Niedermayer 2014-02-14 21:59:56 +01:00
commit f11905763c
1 changed files with 10 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
libavcodec/ppc/dsputil_altivec.c
View File

@ -285,10 +285,10 @@ static int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, in
/* Read potentially unaligned pixels into t1 and t2 /* Read potentially unaligned pixels into t1 and t2
Since we're reading 16 pixels, and actually only want 8, Since we're reading 16 pixels, and actually only want 8,
mask out the last 8 pixels. The 0s don't change the sum. */ mask out the last 8 pixels. The 0s don't change the sum. */
vector unsigned char pix1l = vec_ld( 0, pix1); vector unsigned char pix1l = vec_ld(0, pix1);
vector unsigned char pix1r = vec_ld(15, pix1); vector unsigned char pix1r = vec_ld(7, pix1);
vector unsigned char pix2l = vec_ld( 0, pix2); vector unsigned char pix2l = vec_ld(0, pix2);
vector unsigned char pix2r = vec_ld(15, pix2); vector unsigned char pix2r = vec_ld(7, pix2);
t1 = vec_and(vec_perm(pix1l, pix1r, perm1), permclear); t1 = vec_and(vec_perm(pix1l, pix1r, perm1), permclear);
t2 = vec_and(vec_perm(pix2l, pix2r, perm2), permclear); t2 = vec_and(vec_perm(pix2l, pix2r, perm2), permclear);
@ -367,10 +367,10 @@ static int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, in
/* Read potentially unaligned pixels into t1 and t2 /* Read potentially unaligned pixels into t1 and t2
Since we're reading 16 pixels, and actually only want 8, Since we're reading 16 pixels, and actually only want 8,
mask out the last 8 pixels. The 0s don't change the sum. */ mask out the last 8 pixels. The 0s don't change the sum. */
vector unsigned char pix1l = vec_ld( 0, pix1); vector unsigned char pix1l = vec_ld(0, pix1);
vector unsigned char pix1r = vec_ld(15, pix1); vector unsigned char pix1r = vec_ld(7, pix1);
vector unsigned char pix2l = vec_ld( 0, pix2); vector unsigned char pix2l = vec_ld(0, pix2);
vector unsigned char pix2r = vec_ld(15, pix2); vector unsigned char pix2r = vec_ld(7, pix2);
t1 = vec_and(vec_perm(pix1l, pix1r, perm1), permclear); t1 = vec_and(vec_perm(pix1l, pix1r, perm1), permclear);
t2 = vec_and(vec_perm(pix2l, pix2r, perm2), permclear); t2 = vec_and(vec_perm(pix2l, pix2r, perm2), permclear);
@ -489,8 +489,8 @@ static void get_pixels_altivec(int16_t *restrict block, const uint8_t *pixels, i
// Read potentially unaligned pixels. // Read potentially unaligned pixels.
// We're reading 16 pixels, and actually only want 8, // We're reading 16 pixels, and actually only want 8,
// but we simply ignore the extras. // but we simply ignore the extras.
vector unsigned char pixl = vec_ld( 0, pixels); vector unsigned char pixl = vec_ld(0, pixels);
vector unsigned char pixr = vec_ld(15, pixels); vector unsigned char pixr = vec_ld(7, pixels);
bytes = vec_perm(pixl, pixr, perm); bytes = vec_perm(pixl, pixr, perm);
// convert the bytes into shorts // convert the bytes into shorts