[x265] [PATCH] asm: chroma_hpp[48x64] for i444 - improved 17498c->13381c

aasaipriya at multicorewareinc.com aasaipriya at multicorewareinc.com
Thu Apr 30 05:14:53 CEST 2015


# HG changeset patch
# User Aasaipriya Chandran <aasaipriya at multicorewareinc.com>
# Date 1430307663 -19800
#      Wed Apr 29 17:11:03 2015 +0530
# Node ID 49284d3aae18234c1ea6551ebed834109fe7791a
# Parent  74d7fe7a81ad773c8ae66f076c4e319b253d94ab
asm: chroma_hpp[48x64] for i444 - improved 17498c->13381c

diff -r 74d7fe7a81ad -r 49284d3aae18 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Wed Apr 29 11:08:44 2015 -0500
+++ b/source/common/x86/asm-primitives.cpp	Wed Apr 29 17:11:03 2015 +0530
@@ -2408,6 +2408,7 @@
         p.chroma[X265_CSP_I444].pu[LUMA_64x32].filter_hpp = x265_interp_4tap_horiz_pp_64x32_avx2;
         p.chroma[X265_CSP_I444].pu[LUMA_64x48].filter_hpp = x265_interp_4tap_horiz_pp_64x48_avx2;
         p.chroma[X265_CSP_I444].pu[LUMA_64x16].filter_hpp = x265_interp_4tap_horiz_pp_64x16_avx2;
+        p.chroma[X265_CSP_I444].pu[LUMA_48x64].filter_hpp = x265_interp_4tap_horiz_pp_48x64_avx2;
 
         p.chroma[X265_CSP_I422].pu[CHROMA_422_4x4].filter_hps = x265_interp_4tap_horiz_ps_4x4_avx2;
         p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].filter_hps = x265_interp_4tap_horiz_ps_4x8_avx2;
diff -r 74d7fe7a81ad -r 49284d3aae18 source/common/x86/ipfilter8.asm
--- a/source/common/x86/ipfilter8.asm	Wed Apr 29 11:08:44 2015 -0500
+++ b/source/common/x86/ipfilter8.asm	Wed Apr 29 17:11:03 2015 +0530
@@ -23540,3 +23540,89 @@
     IPFILTER_CHROMA_PP_64xN_AVX2  32
     IPFILTER_CHROMA_PP_64xN_AVX2  48
     IPFILTER_CHROMA_PP_64xN_AVX2  16
+
+;-------------------------------------------------------------------------------------------------------------
+; void interp_4tap_horiz_pp_48x64(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx
+;-------------------------------------------------------------------------------------------------------------
+INIT_YMM avx2
+cglobal interp_4tap_horiz_pp_48x64, 4,6,7
+    mov             r4d, r4m
+
+%ifdef PIC
+    lea               r5,            [tab_ChromaCoeff]
+    vpbroadcastd      m0,            [r5 + r4 * 4]
+%else
+    vpbroadcastd      m0,            [tab_ChromaCoeff + r4 * 4]
+%endif
+
+    mova              m1,            [interp4_horiz_shuf1]
+    vpbroadcastd      m2,            [pw_1]
+    mova              m6,            [pw_512]
+    ; register map
+    ; m0 - interpolate coeff
+    ; m1 - shuffle order table
+    ; m2 - constant word 1
+
+    dec               r0
+    mov               r4d,           64
+
+.loop:
+    ; Row 0
+    vbroadcasti128    m3,            [r0]                        ; [x x x x x A 9 8 7 6 5 4 3 2 1 0]
+    pshufb            m3,            m1
+    pmaddubsw         m3,            m0
+    pmaddwd           m3,            m2
+    vbroadcasti128    m4,            [r0 + 4]
+    pshufb            m4,            m1
+    pmaddubsw         m4,            m0
+    pmaddwd           m4,            m2
+    packssdw          m3,            m4
+    pmulhrsw          m3,            m6
+
+    vbroadcasti128    m4,            [r0 + 16]
+    pshufb            m4,            m1
+    pmaddubsw         m4,            m0
+    pmaddwd           m4,            m2
+    vbroadcasti128    m5,            [r0 + 20]
+    pshufb            m5,            m1
+    pmaddubsw         m5,            m0
+    pmaddwd           m5,            m2
+    packssdw          m4,            m5
+    pmulhrsw          m4,            m6
+
+    packuswb          m3,            m4
+    vpermq            m3,            m3,      q3120
+
+    movu              [r2],          m3
+
+    vbroadcasti128    m3,            [r0 + mmsize]                        ; [x x x x x A 9 8 7 6 5 4 3 2 1 0]
+    pshufb            m3,            m1
+    pmaddubsw         m3,            m0
+    pmaddwd           m3,            m2
+    vbroadcasti128    m4,            [r0 + mmsize + 4]
+    pshufb            m4,            m1
+    pmaddubsw         m4,            m0
+    pmaddwd           m4,            m2
+    packssdw          m3,            m4
+    pmulhrsw          m3,            m6
+
+    vbroadcasti128    m4,            [r0 + mmsize + 16]
+    pshufb            m4,            m1
+    pmaddubsw         m4,            m0
+    pmaddwd           m4,            m2
+    vbroadcasti128    m5,            [r0 + mmsize + 20]
+    pshufb            m5,            m1
+    pmaddubsw         m5,            m0
+    pmaddwd           m5,            m2
+    packssdw          m4,            m5
+    pmulhrsw          m4,            m6
+
+    packuswb          m3,            m4
+    vpermq            m3,            m3,      q3120
+    movu              [r2 + mmsize], xm3
+
+    add               r2,            r3
+    add               r0,            r1
+    dec               r4d
+    jnz               .loop
+    RET


More information about the x265-devel mailing list