[x265] [PATCH 7 of 9] asm: intra_pred_ang8_23 improved by ~18% over SSE4
praveen at multicorewareinc.com
praveen at multicorewareinc.com
Thu Apr 9 15:20:15 CEST 2015
# HG changeset patch
# User Praveen Tiwari <praveen at multicorewareinc.com>
# Date 1428581726 -19800
# Thu Apr 09 17:45:26 2015 +0530
# Node ID 5db8882a1ef6c5f27bc59869692791290945af2a
# Parent 31ce12d63d6560df4ce29bdb948525cf73f057f4
asm: intra_pred_ang8_23 improved by ~18% over SSE4
AVX2:
intra_ang_8x8[23] 9.75x 205.43 2002.05
SSE4:
intra_ang_8x8[23] 8.12x 251.42 2041.61
diff -r 31ce12d63d65 -r 5db8882a1ef6 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp Thu Apr 09 17:08:14 2015 +0530
+++ b/source/common/x86/asm-primitives.cpp Thu Apr 09 17:45:26 2015 +0530
@@ -1765,6 +1765,7 @@
p.cu[BLOCK_8x8].intra_pred[24] = x265_intra_pred_ang8_24_avx2;
p.cu[BLOCK_8x8].intra_pred[11] = x265_intra_pred_ang8_11_avx2;
p.cu[BLOCK_8x8].intra_pred[13] = x265_intra_pred_ang8_13_avx2;
+ p.cu[BLOCK_8x8].intra_pred[23] = x265_intra_pred_ang8_23_avx2;
p.cu[BLOCK_8x8].intra_pred[14] = x265_intra_pred_ang8_14_avx2;
p.cu[BLOCK_8x8].intra_pred[15] = x265_intra_pred_ang8_15_avx2;
p.cu[BLOCK_16x16].intra_pred[3] = x265_intra_pred_ang16_3_avx2;
diff -r 31ce12d63d65 -r 5db8882a1ef6 source/common/x86/intrapred.h
--- a/source/common/x86/intrapred.h Thu Apr 09 17:08:14 2015 +0530
+++ b/source/common/x86/intrapred.h Thu Apr 09 17:45:26 2015 +0530
@@ -236,6 +236,7 @@
void x265_intra_pred_ang8_13_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
void x265_intra_pred_ang8_14_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
void x265_intra_pred_ang8_15_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
+void x265_intra_pred_ang8_23_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
void x265_intra_pred_ang16_3_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
void x265_intra_pred_ang16_4_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
void x265_intra_pred_ang16_5_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
diff -r 31ce12d63d65 -r 5db8882a1ef6 source/common/x86/intrapred8.asm
--- a/source/common/x86/intrapred8.asm Thu Apr 09 17:08:14 2015 +0530
+++ b/source/common/x86/intrapred8.asm Thu Apr 09 17:45:26 2015 +0530
@@ -12056,6 +12056,51 @@
movhps [r0 + r3], xm2
RET
+
+INIT_YMM avx2
+cglobal intra_pred_ang8_23, 3, 6, 6
+ mova m3, [pw_1024]
+ movu xm5, [r2]
+ lea r5, [intra_pred_shuff_0_8]
+ vinserti128 m0, m5, xm5, 1
+ pshufb m0, [r5]
+
+ lea r4, [c_ang8_mode_13]
+ pmaddubsw m1, m0, [r4]
+ pmulhrsw m1, m3
+ pslldq xm5, 1
+ pinsrb xm5, [r2 + 4 + 16], 0
+ pshufb xm4, xm5, [r5]
+ vinserti128 m0, m0, xm4, 1
+ pmaddubsw m2, m0, [r4 + mmsize]
+ pmulhrsw m2, m3
+ vinserti128 m0, m0, xm4, 0
+ pmaddubsw m4, m0, [r4 + 2 * mmsize]
+ pmulhrsw m4, m3
+ pslldq xm5, 1
+ pinsrb xm5, [r2 + 7 + 16], 0
+ pshufb xm5, [r5]
+ vinserti128 m0, m0, xm5, 1
+ pmaddubsw m0, [r4 + 3 * mmsize]
+ pmulhrsw m0, m3
+
+ packuswb m1, m2
+ packuswb m4, m0
+
+ lea r3, [3 * r1]
+ movq [r0], xm1
+ vextracti128 xm2, m1, 1
+ movq [r0 + r1], xm2
+ movhps [r0 + 2 * r1], xm1
+ movhps [r0 + r3], xm2
+ lea r0, [r0 + 4 * r1]
+ movq [r0], xm4
+ vextracti128 xm2, m4, 1
+ movq [r0 + r1], xm2
+ movhps [r0 + 2 * r1], xm4
+ movhps [r0 + r3], xm2
+ RET
+
INIT_YMM avx2
cglobal intra_pred_ang8_12, 3, 5, 5
mova m3, [pw_1024]
More information about the x265-devel
mailing list