Commit 1a5865b6 authored by Martin Vignali's avatar Martin Vignali

avfilter/vf_interlace : add AVX2 for lowpass_line 8 and 16

parent adff97be
...@@ -201,5 +201,10 @@ LOWPASS_LINE ...@@ -201,5 +201,10 @@ LOWPASS_LINE
INIT_XMM avx INIT_XMM avx
LOWPASS_LINE LOWPASS_LINE
%if HAVE_AVX2_EXTERNAL
INIT_YMM avx2
LOWPASS_LINE
%endif
INIT_XMM sse2 INIT_XMM sse2
LOWPASS_LINE_COMPLEX LOWPASS_LINE_COMPLEX
...@@ -32,6 +32,9 @@ void ff_lowpass_line_sse2(uint8_t *dstp, ptrdiff_t linesize, ...@@ -32,6 +32,9 @@ void ff_lowpass_line_sse2(uint8_t *dstp, ptrdiff_t linesize,
void ff_lowpass_line_avx (uint8_t *dstp, ptrdiff_t linesize, void ff_lowpass_line_avx (uint8_t *dstp, ptrdiff_t linesize,
const uint8_t *srcp, ptrdiff_t mref, const uint8_t *srcp, ptrdiff_t mref,
ptrdiff_t pref, int clip_max); ptrdiff_t pref, int clip_max);
void ff_lowpass_line_avx2 (uint8_t *dstp, ptrdiff_t linesize,
const uint8_t *srcp, ptrdiff_t mref,
ptrdiff_t pref, int clip_max);
void ff_lowpass_line_16_sse2(uint8_t *dstp, ptrdiff_t linesize, void ff_lowpass_line_16_sse2(uint8_t *dstp, ptrdiff_t linesize,
const uint8_t *srcp, ptrdiff_t mref, const uint8_t *srcp, ptrdiff_t mref,
...@@ -39,6 +42,9 @@ void ff_lowpass_line_16_sse2(uint8_t *dstp, ptrdiff_t linesize, ...@@ -39,6 +42,9 @@ void ff_lowpass_line_16_sse2(uint8_t *dstp, ptrdiff_t linesize,
void ff_lowpass_line_16_avx (uint8_t *dstp, ptrdiff_t linesize, void ff_lowpass_line_16_avx (uint8_t *dstp, ptrdiff_t linesize,
const uint8_t *srcp, ptrdiff_t mref, const uint8_t *srcp, ptrdiff_t mref,
ptrdiff_t pref, int clip_max); ptrdiff_t pref, int clip_max);
void ff_lowpass_line_16_avx2 (uint8_t *dstp, ptrdiff_t linesize,
const uint8_t *srcp, ptrdiff_t mref,
ptrdiff_t pref, int clip_max);
void ff_lowpass_line_complex_sse2(uint8_t *dstp, ptrdiff_t linesize, void ff_lowpass_line_complex_sse2(uint8_t *dstp, ptrdiff_t linesize,
const uint8_t *srcp, ptrdiff_t mref, const uint8_t *srcp, ptrdiff_t mref,
...@@ -62,6 +68,9 @@ av_cold void ff_interlace_init_x86(InterlaceContext *s, int depth) ...@@ -62,6 +68,9 @@ av_cold void ff_interlace_init_x86(InterlaceContext *s, int depth)
if (EXTERNAL_AVX(cpu_flags)) if (EXTERNAL_AVX(cpu_flags))
if (s->lowpass == VLPF_LIN) if (s->lowpass == VLPF_LIN)
s->lowpass_line = ff_lowpass_line_16_avx; s->lowpass_line = ff_lowpass_line_16_avx;
if (EXTERNAL_AVX2_FAST(cpu_flags))
if (s->lowpass == VLPF_LIN)
s->lowpass_line = ff_lowpass_line_16_avx2;
} else { } else {
if (EXTERNAL_SSE2(cpu_flags)) { if (EXTERNAL_SSE2(cpu_flags)) {
if (s->lowpass == VLPF_LIN) if (s->lowpass == VLPF_LIN)
...@@ -72,5 +81,8 @@ av_cold void ff_interlace_init_x86(InterlaceContext *s, int depth) ...@@ -72,5 +81,8 @@ av_cold void ff_interlace_init_x86(InterlaceContext *s, int depth)
if (EXTERNAL_AVX(cpu_flags)) if (EXTERNAL_AVX(cpu_flags))
if (s->lowpass == VLPF_LIN) if (s->lowpass == VLPF_LIN)
s->lowpass_line = ff_lowpass_line_avx; s->lowpass_line = ff_lowpass_line_avx;
if (EXTERNAL_AVX2_FAST(cpu_flags))
if (s->lowpass == VLPF_LIN)
s->lowpass_line = ff_lowpass_line_avx2;
} }
} }
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment