vp9: add 10/12bpp idct_idct_32x32 sse2 SIMD version.

488fadeb · Ronald S. Bultje · 3d0ca2fe · 488fadeb · 488fadeb
Commit 488fadeb authored Oct 12, 2015 by Ronald S. Bultje
Expand all Hide whitespace changes
Inline Side-by-side

Showing with 488 additions and 44 deletions

vp9dsp_init_16bpp_template.c libavcodec/x86/vp9dsp_init_16bpp_template.c +2 -0

vp9itxfm_16bpp.asm libavcodec/x86/vp9itxfm_16bpp.asm +486 -44

No files found.
--- a/libavcodec/x86/vp9dsp_init_16bpp_template.c
+++ b/libavcodec/x86/vp9dsp_init_16bpp_template.c
@@ -136,6 +136,7 @@ decl_itxfm_func(iadst, idct,  4, BPC, sse2);
 decl_itxfm_func(iadst, iadst, 4, BPC, sse2);
 decl_itxfm_funcs(8, BPC, sse2);
 decl_itxfm_funcs(16, BPC, sse2);
+decl_itxfm_func(idct,  idct, 32, BPC, sse2);
 #endif /* HAVE_YASM */

 av_cold void INIT_FUNC(VP9DSPContext *dsp, int bitexact)
@@ -208,6 +209,7 @@ av_cold void INIT_FUNC(VP9DSPContext *dsp, int bitexact)
 #endif
        init_itx_funcs(TX_8X8, 8, BPC, sse2);
        init_itx_funcs(TX_16X16, 16, BPC, sse2);
+        init_itx_func_one(TX_32X32, idct, idct, 32, BPC, sse2);
    }

    if (EXTERNAL_SSSE3(cpu_flags)) {

--- a/libavcodec/x86/vp9itxfm_16bpp.asm
+++ b/libavcodec/x86/vp9itxfm_16bpp.asm