[x265] [PATCH] asm: avx2 code for pixel_add_ps[64x64] for 10 bpp - 178x

[sumalatha at multicorewareinc.com]

# HG changeset patch
# User Sumalatha Polureddy<sumalatha at multicorewareinc.com>
# Date 1425286505 -19800
# Node ID 8d78c9dc559641fa6140ac44ccc6da8171e7fb14
# Parent  360caa3c6a3d41d2b9adc2c5bc64e2f5d611f52b
asm: avx2 code for pixel_add_ps[64x64] for 10 bpp - 178x

add_ps[64x64]  178.62x          5687.26         1015873.50

diff -r 360caa3c6a3d -r 8d78c9dc5596 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Mon Mar 02 14:22:51 2015 +0530
+++ b/source/common/x86/asm-primitives.cpp	Mon Mar 02 14:25:05 2015 +0530
@@ -1072,6 +1072,7 @@
         p.cu[BLOCK_8x8].add_ps = x265_pixel_add_ps_8x8_avx2;
         p.cu[BLOCK_16x16].add_ps = x265_pixel_add_ps_16x16_avx2;
         p.cu[BLOCK_32x32].add_ps = x265_pixel_add_ps_32x32_avx2;
+        p.cu[BLOCK_64x64].add_ps = x265_pixel_add_ps_64x64_avx2;
 
         p.cu[BLOCK_32x32].ssd_s = x265_pixel_ssd_s_32_avx2;
         p.cu[BLOCK_16x16].sse_ss = x265_pixel_ssd_ss_16x16_avx2;
diff -r 360caa3c6a3d -r 8d78c9dc5596 source/common/x86/pixeladd8.asm
--- a/source/common/x86/pixeladd8.asm	Mon Mar 02 14:22:51 2015 +0530
+++ b/source/common/x86/pixeladd8.asm	Mon Mar 02 14:25:05 2015 +0530
@@ -952,6 +952,43 @@
 
     jnz     .loop
     RET
+
+INIT_YMM avx2
+cglobal pixel_add_ps_64x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m5,     [pw_pixel_max]
+    pxor    m4,     m4
+    mov     r6d,    %2
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
+.loop:
+    movu    m0,     [r2]            ; first 16 of row 0 of src0
+    movu    m2,     [r2 + 32]       ; second 16 of row 0 of src0
+    movu    m1,     [r3]            ; first 16 of row 0 of src1
+    movu    m3,     [r3 + 32]       ; second 16 of row 0 of src1
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+    movu    [r0],       m0          ; first 16 of row 0 of dst
+    movu    [r0 + 32],  m2          ; second 16 of row 0 of dst
+
+    movu    m0,     [r2 + 64]       ; first 16 of row 1 of src0
+    movu    m2,     [r2 + 96]       ; second 16 of row 1 of src0
+    movu    m1,     [r3 + 64]       ; first 16 of row 1 of src1
+    movu    m3,     [r3 + 96]       ; second 16 of row 1 of src1
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+    movu    [r0 + 64],  m0          ; first 16 of row 1 of dst
+    movu    [r0 + 96],  m2          ; second 16 of row 1 of dst
+
+    lea     r2,     [r2 + r4 * 1]
+    lea     r3,     [r3 + r5 * 1]
+    lea     r0,     [r0 + r1 * 1]
+
+    dec     r6d
+    jnz     .loop
+    RET
 %else
 INIT_XMM sse4
 cglobal pixel_add_ps_64x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1