[x265] [PATCH] xDCT4 vector function renamed and code cleanup
praveen at multicorewareinc.com
praveen at multicorewareinc.com
Mon Jul 8 13:30:10 CEST 2013
# HG changeset patch
# User praveentiwari
# Date 1373282998 -19800
# Node ID 185bb07cd150afa6795f75fe0e7a551fb0ba3032
# Parent fb95da9bd85ffe33c1c42958dc7e621053eedba4
xDCT4 vector function renamed and code cleanup
diff -r fb95da9bd85f -r 185bb07cd150 source/common/vec/dct.inc
--- a/source/common/vec/dct.inc Mon Jul 08 16:52:14 2013 +0530
+++ b/source/common/vec/dct.inc Mon Jul 08 16:59:58 2013 +0530
@@ -270,7 +270,7 @@
{ 64, -64, 64, -64, 64, -64, 64, -64 },
{ 36, -83, 36, -83, 36, -83, 36, -83 },
};
-void xDCT4(short *src, int *dst, intptr_t nStride)
+void dct4(short *src, int *dst, intptr_t stride)
{
// Const
__m128i c_1 = _mm_set1_epi32(1);
@@ -281,10 +281,10 @@
__m128i T40, T41, T50, T51, T60, T61, T62, T63, T70, T71, T72, T73;
__m128i T50_, T51_;
- __m128i T10 = _mm_loadl_epi64((__m128i*)&src[0 * nStride]);
- __m128i T11 = _mm_loadl_epi64((__m128i*)&src[1 * nStride]);
- __m128i T12 = _mm_loadl_epi64((__m128i*)&src[2 * nStride]);
- __m128i T13 = _mm_loadl_epi64((__m128i*)&src[3 * nStride]);
+ __m128i T10 = _mm_loadl_epi64((__m128i*)&src[0 * stride]);
+ __m128i T11 = _mm_loadl_epi64((__m128i*)&src[1 * stride]);
+ __m128i T12 = _mm_loadl_epi64((__m128i*)&src[2 * stride]);
+ __m128i T13 = _mm_loadl_epi64((__m128i*)&src[3 * stride]);
T20 = _mm_unpacklo_epi64(T10, T11);
T21 = _mm_unpacklo_epi64(T12, T13);
@@ -4087,7 +4087,7 @@
// TODO: in 16bpp mode, the intermediate must be 32-bits
#if !HIGH_BIT_DEPTH && INSTRSET > 4
p.dct[DST_4x4] = dst4;
- p.dct[DCT_4x4] = xDCT4;
+ p.dct[DCT_4x4] = dct4;
p.dct[DCT_8x8] = xDCT8;
p.dct[DCT_16x16] = xDCT16;
p.dct[DCT_32x32] = xDCT32;
More information about the x265-devel
mailing list