diff options
| author | Graham Booker <gbooker@tamu.edu> | 2007-06-17 09:37:13 +0000 |
|---|---|---|
| committer | Guillaume Poirier <gpoirier@mplayerhq.hu> | 2007-06-17 09:37:13 +0000 |
| commit | 22fa38f0c85fb31cddbb0bc22a2df5953c702b95 (patch) | |
| tree | 451e0b76b342c3bbe98106eb729db7197aec477d | |
| parent | 963eca226e395601f89ee6fd1621e529c9d965f0 (diff) | |
part 2/2 of fixing Altivec-accelerated H264 luma inloop filter
In h264_deblock_q1, the result of the deblock needs to be kept to
be used in future deblocks, so return this value now.
Also change the sign of tc0 vector: It is really a signed value, so
treat it as such until after the >=0 check;
then, at that point, after being masked, it can be treated as unsigned.
Patch by Graham Booker % gbooker A tamu P edu%
Originally committed as revision 9349 to svn://svn.ffmpeg.org/ffmpeg/trunk
| -rw-r--r-- | libavcodec/ppc/h264_altivec.c | 24 |
1 files changed, 15 insertions, 9 deletions
diff --git a/libavcodec/ppc/h264_altivec.c b/libavcodec/ppc/h264_altivec.c index 4fc5538d70..69cbc858da 100644 --- a/libavcodec/ppc/h264_altivec.c +++ b/libavcodec/ppc/h264_altivec.c @@ -740,8 +740,8 @@ static inline vector unsigned char h264_deblock_mask ( register vector unsigned return mask; } -// out: p1 = clip((p2 + ((p0 + q0 + 1) >> 1)) >> 1, p1-tc0, p1+tc0) -static inline void h264_deblock_q1(register vector unsigned char p0, +// out: newp1 = clip((p2 + ((p0 + q0 + 1) >> 1)) >> 1, p1-tc0, p1+tc0) +static inline vector unsigned char h264_deblock_q1(register vector unsigned char p0, register vector unsigned char p1, register vector unsigned char p2, register vector unsigned char q0, @@ -753,6 +753,7 @@ static inline void h264_deblock_q1(register vector unsigned char p0, register vector unsigned char ones; register vector unsigned char max; register vector unsigned char min; + register vector unsigned char newp1; temp = vec_xor(average, p2); average = vec_avg(average, p2); /*avg(p2, avg(p0, q0)) */ @@ -761,8 +762,9 @@ static inline void h264_deblock_q1(register vector unsigned char p0, uncliped = vec_subs(average, temp); /*(p2+((p0+q0+1)>>1))>>1 */ max = vec_adds(p1, tc0); min = vec_subs(p1, tc0); - p1 = vec_max(min, uncliped); - p1 = vec_min(max, p1); + newp1 = vec_max(min, uncliped); + newp1 = vec_min(max, newp1); + return newp1; } #define h264_deblock_p0_q0(p0, p1, q0, q1, tc0masked) { \ @@ -804,9 +806,11 @@ static inline void h264_deblock_q1(register vector unsigned char p0, register vector unsigned char mask; \ register vector unsigned char p1mask; \ register vector unsigned char q1mask; \ - register vector unsigned char tc0vec; \ + register vector char tc0vec; \ register vector unsigned char finaltc0; \ register vector unsigned char tc0masked; \ + register vector unsigned char newp1; \ + register vector unsigned char newq1; \ \ temp[0] = alpha; \ temp[1] = beta; \ @@ -819,24 +823,26 @@ static inline void h264_deblock_q1(register vector unsigned char p0, tc0vec = vec_ld(0, temp); \ tc0vec = vec_mergeh(tc0vec, tc0vec); \ tc0vec = vec_mergeh(tc0vec, tc0vec); \ - mask = vec_and(mask, vec_cmpgt(tc0vec, vec_splat_u8(-1))); /* if tc0[i] >= 0 */ \ - finaltc0 = vec_and(tc0vec, mask); /*tc = tc0[i]*/ \ + mask = vec_and(mask, vec_cmpgt(tc0vec, vec_splat_s8(-1))); /* if tc0[i] >= 0 */ \ + finaltc0 = vec_and((vector unsigned char)tc0vec, mask); /* tc = tc0 */ \ \ p1mask = diff_lt_altivec(p2, p0, betavec); \ p1mask = vec_and(p1mask, mask); /* if( |p2 - p0| < beta) */ \ tc0masked = vec_and(p1mask, tc0vec); \ finaltc0 = vec_sub(finaltc0, p1mask); /* tc++ */ \ - h264_deblock_q1(p0, p1, p2, q0, tc0masked); \ + newp1 = h264_deblock_q1(p0, p1, p2, q0, tc0masked); \ /*end if*/ \ \ q1mask = diff_lt_altivec(q2, q0, betavec); \ q1mask = vec_and(q1mask, mask); /* if ( |q2 - q0| < beta ) */\ tc0masked = vec_and(q1mask, tc0vec); \ finaltc0 = vec_sub(finaltc0, q1mask); /* tc++ */ \ - h264_deblock_q1(p0, q1, q2, q0, tc0masked); \ + newq1 = h264_deblock_q1(p0, q1, q2, q0, tc0masked); \ /*end if*/ \ \ h264_deblock_p0_q0(p0, p1, q0, q1, finaltc0); \ + p1 = newp1; \ + q1 = newq1; \ } static void h264_v_loop_filter_luma_altivec(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) { |