History

avfilter/bwdif: add avx2 filter_line function

James Darnley committed 3 years ago

Commit 073ec3b1 parent a9377232 files · +36 −5Browse files

Showing the changes to libavfilter (2 of 2 files). Show all

Unified · Split

2 files
  1. libavfilter/x86/vf_bwdif.asm +24 −5
  2. libavfilter/x86/vf_bwdif_init.c +12 −0
libavfilter/x86/vf_bwdif.asm+24−5
@@ -26,18 +26,22 @@ · 25 unmodified lines
26 26
27%include "libavutil/x86/x86util.asm"27%include "libavutil/x86/x86util.asm"
28 28
29SECTION_RODATA29SECTION_RODATA 32
30 30
31pw_coefhf: times 4 dw 1016, 557031pw_coefhf: times 8 dw 1016, 5570
32pw_coefhf1: times 8 dw -380132pw_coefhf1: times 16 dw -3801
33pw_coefsp: times 4 dw 5077, -98133pw_coefsp: times 8 dw 5077, -981
34pw_splfdif: times 4 dw -768, 76834pw_splfdif: times 8 dw -768, 768
35 35
36SECTION .text36SECTION .text
37 37
38%macro LOAD8 238%macro LOAD8 2
39 %if mmsize == 32
40 pmovzxbw %1, %2
41 %else
39 movh %1, %242 movh %1, %2
40 punpcklbw %1, m743 punpcklbw %1, m7
44 %endif
41%endmacro45%endmacro
42 46
43%macro LOAD12 247%macro LOAD12 2
@@ -45,8 +49,14 @@ SECTION .text · 1 unmodified line
45%endmacro49%endmacro
46 50
47%macro DISP8 051%macro DISP8 0
52 %if mmsize == 32
53 vextracti128 xm1, m2, 1
54 packuswb xm2, xm1
55 movu [dstq], xm2
56 %else
48 packuswb m2, m257 packuswb m2, m2
49 movh [dstq], m258 movh [dstq], m2
59 %endif
50%endmacro60%endmacro
51 61
52%macro DISP12 062%macro DISP12 0
@@ -244,8 +254,12 @@ cglobal bwdif_filter_line_12bit, 4, 9, 13, 0, dst, prev, cur, next, w, \ · 191 unmodified lines
244 prefs, mrefs, prefs2, mrefs2, \254 prefs, mrefs, prefs2, mrefs2, \
245 prefs3, mrefs3, prefs4, \255 prefs3, mrefs3, prefs4, \
246 mrefs4, parity, clip_max256 mrefs4, parity, clip_max
257 %if mmsize == 32
258 vpbroadcastw m12, WORD clip_maxm
259 %else
247 movd m12, DWORD clip_maxm260 movd m12, DWORD clip_maxm
248 SPLATW m12, m12, 0261 SPLATW m12, m12, 0
262 %endif
249%else263%else
250cglobal bwdif_filter_line_12bit, 4, 6, 8, 80, dst, prev, cur, next, w, \264cglobal bwdif_filter_line_12bit, 4, 6, 8, 80, dst, prev, cur, next, w, \
251 prefs, mrefs, prefs2, mrefs2, \265 prefs, mrefs, prefs2, mrefs2, \
@@ -264,3 +278,8 @@ INIT_XMM ssse3 · 12 unmodified lines
264BWDIF278BWDIF
265INIT_XMM sse2279INIT_XMM sse2
266BWDIF280BWDIF
281
282%if HAVE_AVX2_EXTERNAL && ARCH_X86_64
283INIT_YMM avx2
284BWDIF
285%endif
libavfilter/x86/vf_bwdif_init.c+12−0
@@ -32,6 +32,10 @@ void ff_bwdif_filter_line_ssse3(void *dst, void *prev, void *cur, void *next, · 31 unmodified lines
32 int w, int prefs, int mrefs, int prefs2,32 int w, int prefs, int mrefs, int prefs2,
33 int mrefs2, int prefs3, int mrefs3, int prefs4,33 int mrefs2, int prefs3, int mrefs3, int prefs4,
34 int mrefs4, int parity, int clip_max);34 int mrefs4, int parity, int clip_max);
35void ff_bwdif_filter_line_avx2(void *dst, void *prev, void *cur, void *next,
36 int w, int prefs, int mrefs, int prefs2,
37 int mrefs2, int prefs3, int mrefs3, int prefs4,
38 int mrefs4, int parity, int clip_max);
35 39
36void ff_bwdif_filter_line_12bit_sse2(void *dst, void *prev, void *cur, void *next,40void ff_bwdif_filter_line_12bit_sse2(void *dst, void *prev, void *cur, void *next,
37 int w, int prefs, int mrefs, int prefs2,41 int w, int prefs, int mrefs, int prefs2,
@@ -41,6 +45,10 @@ void ff_bwdif_filter_line_12bit_ssse3(void *dst, void *prev, void *cur, void *ne · 3 unmodified lines
41 int w, int prefs, int mrefs, int prefs2,45 int w, int prefs, int mrefs, int prefs2,
42 int mrefs2, int prefs3, int mrefs3, int prefs4,46 int mrefs2, int prefs3, int mrefs3, int prefs4,
43 int mrefs4, int parity, int clip_max);47 int mrefs4, int parity, int clip_max);
48void ff_bwdif_filter_line_12bit_avx2(void *dst, void *prev, void *cur, void *next,
49 int w, int prefs, int mrefs, int prefs2,
50 int mrefs2, int prefs3, int mrefs3, int prefs4,
51 int mrefs4, int parity, int clip_max);
44 52
45av_cold void ff_bwdif_init_x86(BWDIFContext *bwdif, int bit_depth)53av_cold void ff_bwdif_init_x86(BWDIFContext *bwdif, int bit_depth)
46{54{
@@ -51,10 +59,14 @@ av_cold void ff_bwdif_init_x86(BWDIFContext *bwdif, int bit_depth) · 4 unmodified lines
51 bwdif->filter_line = ff_bwdif_filter_line_sse2;59 bwdif->filter_line = ff_bwdif_filter_line_sse2;
52 if (EXTERNAL_SSSE3(cpu_flags))60 if (EXTERNAL_SSSE3(cpu_flags))
53 bwdif->filter_line = ff_bwdif_filter_line_ssse3;61 bwdif->filter_line = ff_bwdif_filter_line_ssse3;
62 if (ARCH_X86_64 && EXTERNAL_AVX2(cpu_flags))
63 bwdif->filter_line = ff_bwdif_filter_line_avx2;
54 } else if (bit_depth <= 12) {64 } else if (bit_depth <= 12) {
55 if (EXTERNAL_SSE2(cpu_flags))65 if (EXTERNAL_SSE2(cpu_flags))
56 bwdif->filter_line = ff_bwdif_filter_line_12bit_sse2;66 bwdif->filter_line = ff_bwdif_filter_line_12bit_sse2;
57 if (EXTERNAL_SSSE3(cpu_flags))67 if (EXTERNAL_SSSE3(cpu_flags))
58 bwdif->filter_line = ff_bwdif_filter_line_12bit_ssse3;68 bwdif->filter_line = ff_bwdif_filter_line_12bit_ssse3;
69 if (ARCH_X86_64 && EXTERNAL_AVX2(cpu_flags))
70 bwdif->filter_line = ff_bwdif_filter_line_12bit_avx2;
59 }71 }
60}72}

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t