diff options
Diffstat (limited to 'vpx_dsp')
| -rw-r--r-- | vpx_dsp/loopfilter.c | 8 | ||||
| -rw-r--r-- | vpx_dsp/vpx_dsp_rtcd_defs.pl | 2 | ||||
| -rw-r--r-- | vpx_dsp/x86/highbd_loopfilter_sse2.c | 12 |
3 files changed, 9 insertions, 13 deletions
diff --git a/vpx_dsp/loopfilter.c b/vpx_dsp/loopfilter.c index 740d74757..8b740f557 100644 --- a/vpx_dsp/loopfilter.c +++ b/vpx_dsp/loopfilter.c @@ -531,12 +531,12 @@ static INLINE void highbd_filter8(int8_t mask, uint8_t thresh, uint8_t flat, void vpx_highbd_lpf_horizontal_8_c(uint16_t *s, int p, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, - int count, int bd) { + int bd) { int i; // loop filter designed to work using chars so that we can make maximum use // of 8 bit simd instructions. - for (i = 0; i < 8 * count; ++i) { + for (i = 0; i < 8; ++i) { const uint16_t p3 = s[-4 * p], p2 = s[-3 * p], p1 = s[-2 * p], p0 = s[-p]; const uint16_t q0 = s[0 * p], q1 = s[1 * p], q2 = s[2 * p], q3 = s[3 * p]; @@ -559,8 +559,8 @@ void vpx_highbd_lpf_horizontal_8_dual_c(uint16_t *s, int p, const uint8_t *limit1, const uint8_t *thresh1, int bd) { - vpx_highbd_lpf_horizontal_8_c(s, p, blimit0, limit0, thresh0, 1, bd); - vpx_highbd_lpf_horizontal_8_c(s + 8, p, blimit1, limit1, thresh1, 1, bd); + vpx_highbd_lpf_horizontal_8_c(s, p, blimit0, limit0, thresh0, bd); + vpx_highbd_lpf_horizontal_8_c(s + 8, p, blimit1, limit1, thresh1, bd); } void vpx_highbd_lpf_vertical_8_c(uint16_t *s, int pitch, const uint8_t *blimit, diff --git a/vpx_dsp/vpx_dsp_rtcd_defs.pl b/vpx_dsp/vpx_dsp_rtcd_defs.pl index c2b042271..1e7800ad7 100644 --- a/vpx_dsp/vpx_dsp_rtcd_defs.pl +++ b/vpx_dsp/vpx_dsp_rtcd_defs.pl @@ -587,7 +587,7 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") { add_proto qw/void vpx_highbd_lpf_horizontal_16/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd"; specialize qw/vpx_highbd_lpf_horizontal_16 sse2/; - add_proto qw/void vpx_highbd_lpf_horizontal_8/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd"; + add_proto qw/void vpx_highbd_lpf_horizontal_8/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd"; specialize qw/vpx_highbd_lpf_horizontal_8 sse2/; add_proto qw/void vpx_highbd_lpf_horizontal_8_dual/, "uint16_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1, int bd"; diff --git a/vpx_dsp/x86/highbd_loopfilter_sse2.c b/vpx_dsp/x86/highbd_loopfilter_sse2.c index a48bbd316..53786de83 100644 --- a/vpx_dsp/x86/highbd_loopfilter_sse2.c +++ b/vpx_dsp/x86/highbd_loopfilter_sse2.c @@ -523,7 +523,7 @@ void vpx_highbd_lpf_horizontal_8_sse2(uint16_t *s, int p, const uint8_t *_blimit, const uint8_t *_limit, const uint8_t *_thresh, - int count, int bd) { + int bd) { DECLARE_ALIGNED(16, uint16_t, flat_op2[16]); DECLARE_ALIGNED(16, uint16_t, flat_op1[16]); DECLARE_ALIGNED(16, uint16_t, flat_op0[16]); @@ -556,8 +556,6 @@ void vpx_highbd_lpf_horizontal_8_sse2(uint16_t *s, int p, __m128i work_a; __m128i filter1, filter2; - (void)count; - if (bd == 8) { blimit = _mm_unpacklo_epi8(_mm_load_si128((const __m128i *)_blimit), zero); limit = _mm_unpacklo_epi8(_mm_load_si128((const __m128i *)_limit), zero); @@ -764,9 +762,8 @@ void vpx_highbd_lpf_horizontal_8_dual_sse2(uint16_t *s, int p, const uint8_t *_limit1, const uint8_t *_thresh1, int bd) { - vpx_highbd_lpf_horizontal_8_sse2(s, p, _blimit0, _limit0, _thresh0, 1, bd); - vpx_highbd_lpf_horizontal_8_sse2(s + 8, p, _blimit1, _limit1, _thresh1, - 1, bd); + vpx_highbd_lpf_horizontal_8_sse2(s, p, _blimit0, _limit0, _thresh0, bd); + vpx_highbd_lpf_horizontal_8_sse2(s + 8, p, _blimit1, _limit1, _thresh1, bd); } void vpx_highbd_lpf_horizontal_4_sse2(uint16_t *s, int p, @@ -1123,8 +1120,7 @@ void vpx_highbd_lpf_vertical_8_sse2(uint16_t *s, int p, highbd_transpose(src, p, dst, 8, 1); // Loop filtering - vpx_highbd_lpf_horizontal_8_sse2(t_dst + 4 * 8, 8, blimit, limit, thresh, 1, - bd); + vpx_highbd_lpf_horizontal_8_sse2(t_dst + 4 * 8, 8, blimit, limit, thresh, bd); src[0] = t_dst; dst[0] = s - 4; |