ref: 7f1f35183ab26dcedc208b76507f7ffd8207651c
parent: 691ef20272d398c5802b185859a797af3ce1e289
author: Linfeng Zhang <[email protected]>
date: Wed Sep 28 11:20:01 EDT 2016
Unify loopfilter function names Rename vpx_lpf_horizontal_edge_8() to vpx_lpf_horizontal_16(). Rename vpx_lpf_horizontal_edge_16() to vpx_lpf_horizontal_16_dual(). Change-Id: I798ca8fbbd657d06d3db2bfb0fb3321168f49e52
--- a/test/lpf_test.cc
+++ b/test/lpf_test.cc
@@ -402,10 +402,10 @@
&vpx_highbd_lpf_vertical_4_c, 8),
make_tuple(&vpx_highbd_lpf_horizontal_8_sse2,
&vpx_highbd_lpf_horizontal_8_c, 8),
- make_tuple(&vpx_highbd_lpf_horizontal_edge_8_sse2,
- &vpx_highbd_lpf_horizontal_edge_8_c, 8),
- make_tuple(&vpx_highbd_lpf_horizontal_edge_16_sse2,
- &vpx_highbd_lpf_horizontal_edge_16_c, 8),
+ make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+ &vpx_highbd_lpf_horizontal_16_c, 8),
+ make_tuple(&vpx_highbd_lpf_horizontal_16_dual_sse2,
+ &vpx_highbd_lpf_horizontal_16_dual_c, 8),
make_tuple(&vpx_highbd_lpf_vertical_8_sse2,
&vpx_highbd_lpf_vertical_8_c, 8),
make_tuple(&vpx_highbd_lpf_vertical_16_sse2,
@@ -416,10 +416,10 @@
&vpx_highbd_lpf_vertical_4_c, 10),
make_tuple(&vpx_highbd_lpf_horizontal_8_sse2,
&vpx_highbd_lpf_horizontal_8_c, 10),
- make_tuple(&vpx_highbd_lpf_horizontal_edge_8_sse2,
- &vpx_highbd_lpf_horizontal_edge_8_c, 10),
- make_tuple(&vpx_highbd_lpf_horizontal_edge_16_sse2,
- &vpx_highbd_lpf_horizontal_edge_16_c, 10),
+ make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+ &vpx_highbd_lpf_horizontal_16_c, 10),
+ make_tuple(&vpx_highbd_lpf_horizontal_16_dual_sse2,
+ &vpx_highbd_lpf_horizontal_16_dual_c, 10),
make_tuple(&vpx_highbd_lpf_vertical_8_sse2,
&vpx_highbd_lpf_vertical_8_c, 10),
make_tuple(&vpx_highbd_lpf_vertical_16_sse2,
@@ -430,10 +430,10 @@
&vpx_highbd_lpf_vertical_4_c, 12),
make_tuple(&vpx_highbd_lpf_horizontal_8_sse2,
&vpx_highbd_lpf_horizontal_8_c, 12),
- make_tuple(&vpx_highbd_lpf_horizontal_edge_8_sse2,
- &vpx_highbd_lpf_horizontal_edge_8_c, 12),
- make_tuple(&vpx_highbd_lpf_horizontal_edge_16_sse2,
- &vpx_highbd_lpf_horizontal_edge_16_c, 12),
+ make_tuple(&vpx_highbd_lpf_horizontal_16_sse2,
+ &vpx_highbd_lpf_horizontal_16_c, 12),
+ make_tuple(&vpx_highbd_lpf_horizontal_16_dual_sse2,
+ &vpx_highbd_lpf_horizontal_16_dual_c, 12),
make_tuple(&vpx_highbd_lpf_vertical_8_sse2,
&vpx_highbd_lpf_vertical_8_c, 12),
make_tuple(&vpx_highbd_lpf_vertical_16_sse2,
@@ -450,10 +450,9 @@
::testing::Values(
make_tuple(&vpx_lpf_horizontal_4_sse2, &vpx_lpf_horizontal_4_c, 8),
make_tuple(&vpx_lpf_horizontal_8_sse2, &vpx_lpf_horizontal_8_c, 8),
- make_tuple(&vpx_lpf_horizontal_edge_8_sse2,
- &vpx_lpf_horizontal_edge_8_c, 8),
- make_tuple(&vpx_lpf_horizontal_edge_16_sse2,
- &vpx_lpf_horizontal_edge_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_sse2, &vpx_lpf_horizontal_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_dual_sse2,
+ &vpx_lpf_horizontal_16_dual_c, 8),
make_tuple(&vpx_lpf_vertical_4_sse2, &vpx_lpf_vertical_4_c, 8),
make_tuple(&vpx_lpf_vertical_8_sse2, &vpx_lpf_vertical_8_c, 8),
make_tuple(&vpx_lpf_vertical_16_sse2, &vpx_lpf_vertical_16_c, 8),
@@ -465,10 +464,10 @@
#if HAVE_AVX2 && (!CONFIG_VP9_HIGHBITDEPTH)
INSTANTIATE_TEST_CASE_P(
AVX2, Loop8Test6Param,
- ::testing::Values(make_tuple(&vpx_lpf_horizontal_edge_8_avx2,
- &vpx_lpf_horizontal_edge_8_c, 8),
- make_tuple(&vpx_lpf_horizontal_edge_16_avx2,
- &vpx_lpf_horizontal_edge_16_c, 8)));
+ ::testing::Values(make_tuple(&vpx_lpf_horizontal_16_avx2,
+ &vpx_lpf_horizontal_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_dual_avx2,
+ &vpx_lpf_horizontal_16_dual_c, 8)));
#endif
#if HAVE_SSE2
@@ -520,10 +519,9 @@
INSTANTIATE_TEST_CASE_P(
NEON, Loop8Test6Param,
::testing::Values(
- make_tuple(&vpx_lpf_horizontal_edge_8_neon,
- &vpx_lpf_horizontal_edge_8_c, 8),
- make_tuple(&vpx_lpf_horizontal_edge_16_neon,
- &vpx_lpf_horizontal_edge_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_neon, &vpx_lpf_horizontal_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_dual_neon,
+ &vpx_lpf_horizontal_16_dual_c, 8),
make_tuple(&vpx_lpf_vertical_16_neon, &vpx_lpf_vertical_16_c, 8),
make_tuple(&vpx_lpf_vertical_16_dual_neon, &vpx_lpf_vertical_16_dual_c,
8),
@@ -550,8 +548,9 @@
::testing::Values(
make_tuple(&vpx_lpf_horizontal_4_dspr2, &vpx_lpf_horizontal_4_c, 8),
make_tuple(&vpx_lpf_horizontal_8_dspr2, &vpx_lpf_horizontal_8_c, 8),
- make_tuple(&vpx_lpf_horizontal_edge_8, &vpx_lpf_horizontal_edge_8, 8),
- make_tuple(&vpx_lpf_horizontal_edge_16, &vpx_lpf_horizontal_edge_16, 8),
+ make_tuple(&vpx_lpf_horizontal_16_dspr2, &vpx_lpf_horizontal_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_dual_dspr2,
+ &vpx_lpf_horizontal_16_dual_c, 8),
make_tuple(&vpx_lpf_vertical_4_dspr2, &vpx_lpf_vertical_4_c, 8),
make_tuple(&vpx_lpf_vertical_8_dspr2, &vpx_lpf_vertical_8_c, 8),
make_tuple(&vpx_lpf_vertical_16_dspr2, &vpx_lpf_vertical_16_c, 8),
@@ -576,10 +575,9 @@
::testing::Values(
make_tuple(&vpx_lpf_horizontal_4_msa, &vpx_lpf_horizontal_4_c, 8),
make_tuple(&vpx_lpf_horizontal_8_msa, &vpx_lpf_horizontal_8_c, 8),
- make_tuple(&vpx_lpf_horizontal_edge_8_msa, &vpx_lpf_horizontal_edge_8_c,
- 8),
- make_tuple(&vpx_lpf_horizontal_edge_16_msa,
- &vpx_lpf_horizontal_edge_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_msa, &vpx_lpf_horizontal_16_c, 8),
+ make_tuple(&vpx_lpf_horizontal_16_dual_msa,
+ &vpx_lpf_horizontal_16_dual_c, 8),
make_tuple(&vpx_lpf_vertical_4_msa, &vpx_lpf_vertical_4_c, 8),
make_tuple(&vpx_lpf_vertical_8_msa, &vpx_lpf_vertical_8_c, 8),
make_tuple(&vpx_lpf_vertical_16_msa, &vpx_lpf_vertical_16_c, 8)));
--- a/vp9/common/vp9_loopfilter.c
+++ b/vp9/common/vp9_loopfilter.c
@@ -465,12 +465,11 @@
if (mask_16x16 & 1) { if ((mask_16x16 & 3) == 3) {- vpx_lpf_horizontal_edge_16(s, pitch, lfi->mblim, lfi->lim,
+ vpx_lpf_horizontal_16_dual(s, pitch, lfi->mblim, lfi->lim,
lfi->hev_thr);
count = 2;
} else {- vpx_lpf_horizontal_edge_8(s, pitch, lfi->mblim, lfi->lim,
- lfi->hev_thr);
+ vpx_lpf_horizontal_16(s, pitch, lfi->mblim, lfi->lim, lfi->hev_thr);
}
} else if (mask_8x8 & 1) { if ((mask_8x8 & 3) == 3) {@@ -559,12 +558,12 @@
if (mask_16x16 & 1) { if ((mask_16x16 & 3) == 3) {- vpx_highbd_lpf_horizontal_edge_16(s, pitch, lfi->mblim, lfi->lim,
+ vpx_highbd_lpf_horizontal_16_dual(s, pitch, lfi->mblim, lfi->lim,
lfi->hev_thr, bd);
count = 2;
} else {- vpx_highbd_lpf_horizontal_edge_8(s, pitch, lfi->mblim, lfi->lim,
- lfi->hev_thr, bd);
+ vpx_highbd_lpf_horizontal_16(s, pitch, lfi->mblim, lfi->lim,
+ lfi->hev_thr, bd);
}
} else if (mask_8x8 & 1) { if ((mask_8x8 & 3) == 3) {--- a/vpx_dsp/arm/loopfilter_mb_neon.asm
+++ b/vpx_dsp/arm/loopfilter_mb_neon.asm
@@ -8,8 +8,8 @@
; be found in the AUTHORS file in the root of the source tree.
;
- EXPORT |vpx_lpf_horizontal_edge_8_neon|
- EXPORT |vpx_lpf_horizontal_edge_16_neon|
+ EXPORT |vpx_lpf_horizontal_16_neon|
+ EXPORT |vpx_lpf_horizontal_16_dual_neon|
EXPORT |vpx_lpf_vertical_16_neon|
EXPORT |vpx_lpf_vertical_16_dual_neon|
ARM
@@ -119,7 +119,7 @@
ENDP ; |mb_lpf_horizontal_edge|
-; void vpx_lpf_horizontal_edge_8_neon(uint8_t *s, int pitch,
+; void vpx_lpf_horizontal_16_neon(uint8_t *s, int pitch,
; const uint8_t *blimit,
; const uint8_t *limit,
; const uint8_t *thresh)
@@ -128,12 +128,12 @@
; r2 const uint8_t *blimit,
; r3 const uint8_t *limit,
; sp const uint8_t *thresh
-|vpx_lpf_horizontal_edge_8_neon| PROC
+|vpx_lpf_horizontal_16_neon| PROC
mov r12, #1
b mb_lpf_horizontal_edge
- ENDP ; |vpx_lpf_horizontal_edge_8_neon|
+ ENDP ; |vpx_lpf_horizontal_16_neon|
-; void vpx_lpf_horizontal_edge_16_neon(uint8_t *s, int pitch,
+; void vpx_lpf_horizontal_16_dual_neon(uint8_t *s, int pitch,
; const uint8_t *blimit,
; const uint8_t *limit,
; const uint8_t *thresh)
@@ -142,10 +142,10 @@
; r2 const uint8_t *blimit,
; r3 const uint8_t *limit,
; sp const uint8_t *thresh
-|vpx_lpf_horizontal_edge_16_neon| PROC
+|vpx_lpf_horizontal_16_dual_neon| PROC
mov r12, #2
b mb_lpf_horizontal_edge
- ENDP ; |vpx_lpf_horizontal_edge_16_neon|
+ ENDP ; |vpx_lpf_horizontal_16_dual_neon|
; void mb_lpf_vertical_edge_w(uint8_t *s, int p, const uint8_t *blimit,
; const uint8_t *limit, const uint8_t *thresh,
--- a/vpx_dsp/arm/loopfilter_mb_neon.c
+++ b/vpx_dsp/arm/loopfilter_mb_neon.c
@@ -975,9 +975,8 @@
FUN_LPF_16_KERNEL(_dual_, 16) // lpf_16_dual_kernel
#undef FUN_LPF_16_KERNEL
-void vpx_lpf_horizontal_edge_8_neon(uint8_t *s, int p, const uint8_t *blimit,
- const uint8_t *limit,
- const uint8_t *thresh) {+void vpx_lpf_horizontal_16_neon(uint8_t *s, int p, const uint8_t *blimit,
+ const uint8_t *limit, const uint8_t *thresh) {uint8x8_t p7, p6, p5, p4, p3, p2, p1, p0, q0, q1, q2, q3, q4, q5, q6, q7, op6,
op5, op4, op3, op2, op1, op0, oq0, oq1, oq2, oq3, oq4, oq5, oq6;
uint32_t flat_status, flat2_status;
@@ -992,7 +991,7 @@
oq5, oq6, flat_status, flat2_status);
}
-void vpx_lpf_horizontal_edge_16_neon(uint8_t *s, int p, const uint8_t *blimit,
+void vpx_lpf_horizontal_16_dual_neon(uint8_t *s, int p, const uint8_t *blimit,
const uint8_t *limit,
const uint8_t *thresh) {uint8x16_t p7, p6, p5, p4, p3, p2, p1, p0, q0, q1, q2, q3, q4, q5, q6, q7,
--- a/vpx_dsp/loopfilter.c
+++ b/vpx_dsp/loopfilter.c
@@ -308,12 +308,12 @@
}
}
-void vpx_lpf_horizontal_edge_8_c(uint8_t *s, int p, const uint8_t *blimit,
- const uint8_t *limit, const uint8_t *thresh) {+void vpx_lpf_horizontal_16_c(uint8_t *s, int p, const uint8_t *blimit,
+ const uint8_t *limit, const uint8_t *thresh) {mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 1);
}
-void vpx_lpf_horizontal_edge_16_c(uint8_t *s, int p, const uint8_t *blimit,
+void vpx_lpf_horizontal_16_dual_c(uint8_t *s, int p, const uint8_t *blimit,
const uint8_t *limit, const uint8_t *thresh) {mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 2);
}
@@ -673,14 +673,13 @@
}
}
-void vpx_highbd_lpf_horizontal_edge_8_c(uint16_t *s, int p,
- const uint8_t *blimit,
- const uint8_t *limit,
- const uint8_t *thresh, int bd) {+void vpx_highbd_lpf_horizontal_16_c(uint16_t *s, int p, const uint8_t *blimit,
+ const uint8_t *limit, const uint8_t *thresh,
+ int bd) {highbd_mb_lpf_horizontal_edge_w(s, p, blimit, limit, thresh, 1, bd);
}
-void vpx_highbd_lpf_horizontal_edge_16_c(uint16_t *s, int p,
+void vpx_highbd_lpf_horizontal_16_dual_c(uint16_t *s, int p,
const uint8_t *blimit,
const uint8_t *limit,
const uint8_t *thresh, int bd) {--- a/vpx_dsp/mips/loopfilter_16_msa.c
+++ b/vpx_dsp/mips/loopfilter_16_msa.c
@@ -403,10 +403,11 @@
}
}
-void vpx_lpf_horizontal_16_dual_msa(uint8_t *src, int32_t pitch,
- const uint8_t *b_limit_ptr,
- const uint8_t *limit_ptr,
- const uint8_t *thresh_ptr, int32_t count) {+static void mb_lpf_horizontal_edge_dual(uint8_t *src, int32_t pitch,
+ const uint8_t *b_limit_ptr,
+ const uint8_t *limit_ptr,
+ const uint8_t *thresh_ptr,
+ int32_t count) {DECLARE_ALIGNED(32, uint8_t, filter48[16 * 8]);
uint8_t early_exit = 0;
@@ -638,19 +639,19 @@
}
}
} else {- vpx_lpf_horizontal_16_dual_msa(src, pitch, b_limit_ptr, limit_ptr,
- thresh_ptr, count);
+ mb_lpf_horizontal_edge_dual(src, pitch, b_limit_ptr, limit_ptr, thresh_ptr,
+ count);
}
}
-void vpx_lpf_horizontal_edge_8_msa(uint8_t *src, int32_t pitch,
- const uint8_t *b_limit_ptr,
- const uint8_t *limit_ptr,
- const uint8_t *thresh_ptr) {+void vpx_lpf_horizontal_16_msa(uint8_t *src, int32_t pitch,
+ const uint8_t *b_limit_ptr,
+ const uint8_t *limit_ptr,
+ const uint8_t *thresh_ptr) {mb_lpf_horizontal_edge(src, pitch, b_limit_ptr, limit_ptr, thresh_ptr, 1);
}
-void vpx_lpf_horizontal_edge_16_msa(uint8_t *src, int32_t pitch,
+void vpx_lpf_horizontal_16_dual_msa(uint8_t *src, int32_t pitch,
const uint8_t *b_limit_ptr,
const uint8_t *limit_ptr,
const uint8_t *thresh_ptr) {--- a/vpx_dsp/mips/loopfilter_mb_horiz_dspr2.c
+++ b/vpx_dsp/mips/loopfilter_mb_horiz_dspr2.c
@@ -717,14 +717,13 @@
}
}
-void vpx_lpf_horizontal_edge_8_dspr2(unsigned char *s, int pitch,
- const uint8_t *blimit,
- const uint8_t *limit,
- const uint8_t *thresh) {+void vpx_lpf_horizontal_16_dspr2(unsigned char *s, int pitch,
+ const uint8_t *blimit, const uint8_t *limit,
+ const uint8_t *thresh) {mb_lpf_horizontal_edge(s, pitch, blimit, limit, thresh, 1);
}
-void vpx_lpf_horizontal_edge_16_dspr2(unsigned char *s, int pitch,
+void vpx_lpf_horizontal_16_dual_dspr2(unsigned char *s, int pitch,
const uint8_t *blimit,
const uint8_t *limit,
const uint8_t *thresh) {--- a/vpx_dsp/vpx_dsp_rtcd_defs.pl
+++ b/vpx_dsp/vpx_dsp_rtcd_defs.pl
@@ -522,11 +522,11 @@
add_proto qw/void vpx_lpf_vertical_4_dual/, "uint8_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";
specialize qw/vpx_lpf_vertical_4_dual sse2 neon dspr2 msa/;
-add_proto qw/void vpx_lpf_horizontal_edge_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
-specialize qw/vpx_lpf_horizontal_edge_8 sse2 avx2 neon dspr2 msa/;
+add_proto qw/void vpx_lpf_horizontal_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
+specialize qw/vpx_lpf_horizontal_16 sse2 avx2 neon dspr2 msa/;
-add_proto qw/void vpx_lpf_horizontal_edge_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
-specialize qw/vpx_lpf_horizontal_edge_16 sse2 avx2 neon dspr2 msa/;
+add_proto qw/void vpx_lpf_horizontal_16_dual/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
+specialize qw/vpx_lpf_horizontal_16_dual sse2 avx2 neon dspr2 msa/;
add_proto qw/void vpx_lpf_horizontal_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
specialize qw/vpx_lpf_horizontal_8 sse2 neon dspr2 msa/;
@@ -559,11 +559,11 @@
add_proto qw/void vpx_highbd_lpf_vertical_4_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";
specialize qw/vpx_highbd_lpf_vertical_4_dual sse2/;
- add_proto qw/void vpx_highbd_lpf_horizontal_edge_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_edge_8 sse2/;
+ 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 bd";
+ specialize qw/vpx_highbd_lpf_horizontal_16 sse2/;
- add_proto qw/void vpx_highbd_lpf_horizontal_edge_16/, "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_edge_16 sse2/;
+ add_proto qw/void vpx_highbd_lpf_horizontal_16_dual/, "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_16_dual 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 bd";
specialize qw/vpx_highbd_lpf_horizontal_8 sse2/;
--- a/vpx_dsp/x86/highbd_loopfilter_sse2.c
+++ b/vpx_dsp/x86/highbd_loopfilter_sse2.c
@@ -48,10 +48,10 @@
// TODO(debargha, peter): Break up large functions into smaller ones
// in this file.
-void vpx_highbd_lpf_horizontal_edge_8_sse2(uint16_t *s, int p,
- const uint8_t *_blimit,
- const uint8_t *_limit,
- const uint8_t *_thresh, int bd) {+void vpx_highbd_lpf_horizontal_16_sse2(uint16_t *s, int p,
+ const uint8_t *_blimit,
+ const uint8_t *_limit,
+ const uint8_t *_thresh, int bd) {const __m128i zero = _mm_set1_epi16(0);
const __m128i one = _mm_set1_epi16(1);
__m128i blimit, limit, thresh;
@@ -475,12 +475,12 @@
_mm_store_si128((__m128i *)(s - 0 * p), q0);
}
-void vpx_highbd_lpf_horizontal_edge_16_sse2(uint16_t *s, int p,
+void vpx_highbd_lpf_horizontal_16_dual_sse2(uint16_t *s, int p,
const uint8_t *_blimit,
const uint8_t *_limit,
const uint8_t *_thresh, int bd) {- vpx_highbd_lpf_horizontal_edge_8_sse2(s, p, _blimit, _limit, _thresh, bd);
- vpx_highbd_lpf_horizontal_edge_8_sse2(s + 8, p, _blimit, _limit, _thresh, bd);
+ vpx_highbd_lpf_horizontal_16_sse2(s, p, _blimit, _limit, _thresh, bd);
+ vpx_highbd_lpf_horizontal_16_sse2(s + 8, p, _blimit, _limit, _thresh, bd);
}
void vpx_highbd_lpf_horizontal_8_sse2(uint16_t *s, int p,
@@ -1108,8 +1108,8 @@
highbd_transpose(src, p, dst, 8, 2);
// Loop filtering
- vpx_highbd_lpf_horizontal_edge_8_sse2(t_dst + 8 * 8, 8, blimit, limit, thresh,
- bd);
+ vpx_highbd_lpf_horizontal_16_sse2(t_dst + 8 * 8, 8, blimit, limit, thresh,
+ bd);
src[0] = t_dst;
src[1] = t_dst + 8 * 8;
dst[0] = s - 8;
@@ -1130,7 +1130,7 @@
highbd_transpose8x16(s, s + 8 * p, p, t_dst + 8 * 16, 16);
// Loop filtering
- vpx_highbd_lpf_horizontal_edge_16_sse2(t_dst + 8 * 16, 16, blimit, limit,
+ vpx_highbd_lpf_horizontal_16_dual_sse2(t_dst + 8 * 16, 16, blimit, limit,
thresh, bd);
// Transpose back
--- a/vpx_dsp/x86/loopfilter_avx2.c
+++ b/vpx_dsp/x86/loopfilter_avx2.c
@@ -13,10 +13,10 @@
#include "./vpx_dsp_rtcd.h"
#include "vpx_ports/mem.h"
-void vpx_lpf_horizontal_edge_8_avx2(unsigned char *s, int p,
- const unsigned char *_blimit,
- const unsigned char *_limit,
- const unsigned char *_thresh) {+void vpx_lpf_horizontal_16_avx2(unsigned char *s, int p,
+ const unsigned char *_blimit,
+ const unsigned char *_limit,
+ const unsigned char *_thresh) {__m128i mask, hev, flat, flat2;
const __m128i zero = _mm_set1_epi16(0);
const __m128i one = _mm_set1_epi8(1);
@@ -367,7 +367,7 @@
8, 128, 9, 128, 10, 128, 11, 128, 12, 128, 13, 128, 14, 128, 15, 128
};
-void vpx_lpf_horizontal_edge_16_avx2(unsigned char *s, int p,
+void vpx_lpf_horizontal_16_dual_avx2(unsigned char *s, int p,
const unsigned char *_blimit,
const unsigned char *_limit,
const unsigned char *_thresh) {--- a/vpx_dsp/x86/loopfilter_sse2.c
+++ b/vpx_dsp/x86/loopfilter_sse2.c
@@ -229,10 +229,10 @@
*(int *)(s + 7 * p - 2) = _mm_cvtsi128_si32(qs1qs0);
}
-void vpx_lpf_horizontal_edge_8_sse2(unsigned char *s, int p,
- const unsigned char *_blimit,
- const unsigned char *_limit,
- const unsigned char *_thresh) {+void vpx_lpf_horizontal_16_sse2(unsigned char *s, int p,
+ const unsigned char *_blimit,
+ const unsigned char *_limit,
+ const unsigned char *_thresh) {const __m128i zero = _mm_set1_epi16(0);
const __m128i one = _mm_set1_epi8(1);
const __m128i blimit = _mm_load_si128((const __m128i *)_blimit);
@@ -591,7 +591,7 @@
return _mm_or_si128(_mm_andnot_si128(*flat, *other_filt), result);
}
-void vpx_lpf_horizontal_edge_16_sse2(unsigned char *s, int p,
+void vpx_lpf_horizontal_16_dual_sse2(unsigned char *s, int p,
const unsigned char *_blimit,
const unsigned char *_limit,
const unsigned char *_thresh) {@@ -1745,7 +1745,7 @@
transpose(src, p, dst, 8, 2);
// Loop filtering
- vpx_lpf_horizontal_edge_8_sse2(t_dst + 8 * 8, 8, blimit, limit, thresh);
+ vpx_lpf_horizontal_16_sse2(t_dst + 8 * 8, 8, blimit, limit, thresh);
src[0] = t_dst;
src[1] = t_dst + 8 * 8;
@@ -1766,7 +1766,7 @@
transpose8x16(s, s + 8 * p, p, t_dst + 8 * 16, 16);
// Loop filtering
- vpx_lpf_horizontal_edge_16_sse2(t_dst + 8 * 16, 16, blimit, limit, thresh);
+ vpx_lpf_horizontal_16_dual_sse2(t_dst + 8 * 16, 16, blimit, limit, thresh);
// Transpose back
transpose8x16(t_dst, t_dst + 8 * 16, 16, s - 8, p);