swscale: move yuv2yuvX_sse3 to yasm, unrolls main loop

And other small optimizations for ~20% speedup.

4 files changed, 191 insertions, 158 deletions

diff --git a/libswscale/x86/Makefile b/libswscale/x86/Makefile
index 831d5359aa..bfe383364e 100644
--- a/libswscale/x86/Makefile
+++ b/libswscale/x86/Makefile
@@ -13,3 +13,4 @@ X86ASM-OBJS                     += x86/input.o                          \
                                    x86/scale.o                          \
                                    x86/rgb_2_rgb.o                      \
                                    x86/yuv_2_rgb.o                      \
+                                   x86/yuv2yuvX.o                       \
diff --git a/libswscale/x86/swscale.c b/libswscale/x86/swscale.c
index 15c0b22f20..3df193a067 100644
--- a/libswscale/x86/swscale.c
+++ b/libswscale/x86/swscale.c
@@ -63,6 +63,16 @@ DECLARE_ASM_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
 DECLARE_ASM_ALIGNED(8, const uint64_t, ff_w1111)        = 0x0001000100010001ULL;
 
 
+#define YUV2YUVX_FUNC_DECL(opt)  \
+static void yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, const int16_t **src, \
+                           uint8_t *dest, int dstW, \
+                           const uint8_t *dither, int offset); \
+
+YUV2YUVX_FUNC_DECL(mmx)
+YUV2YUVX_FUNC_DECL(mmxext)
+YUV2YUVX_FUNC_DECL(sse3)
+YUV2YUVX_FUNC_DECL(avx2)
+
 //MMX versions
 #if HAVE_MMX_INLINE
 #undef RENAME
@@ -198,81 +208,44 @@ void ff_updateMMXDitherTables(SwsContext *c, int dstY)
 }
 
 #if HAVE_MMXEXT
-static void yuv2yuvX_sse3(const int16_t *filter, int filterSize,
-                           const int16_t **src, uint8_t *dest, int dstW,
-                           const uint8_t *dither, int offset)
-{
-    if(((uintptr_t)dest) & 15){
-        yuv2yuvX_mmxext(filter, filterSize, src, dest, dstW, dither, offset);
-        return;
-    }
-    filterSize--;
-#define MAIN_FUNCTION \
-        "pxor       %%xmm0, %%xmm0 \n\t" \
-        "punpcklbw  %%xmm0, %%xmm3 \n\t" \
-        "movd           %4, %%xmm1 \n\t" \
-        "punpcklwd  %%xmm1, %%xmm1 \n\t" \
-        "punpckldq  %%xmm1, %%xmm1 \n\t" \
-        "punpcklqdq %%xmm1, %%xmm1 \n\t" \
-        "psllw          $3, %%xmm1 \n\t" \
-        "paddw      %%xmm1, %%xmm3 \n\t" \
-        "psraw          $4, %%xmm3 \n\t" \
-        "movdqa     %%xmm3, %%xmm4 \n\t" \
-        "movdqa     %%xmm3, %%xmm7 \n\t" \
-        "movl           %3, %%ecx  \n\t" \
-        "mov                                 %0, %%"FF_REG_d"        \n\t"\
-        "mov                        (%%"FF_REG_d"), %%"FF_REG_S"     \n\t"\
-        ".p2align                             4             \n\t" /* FIXME Unroll? */\
-        "1:                                                 \n\t"\
-        "movddup                  8(%%"FF_REG_d"), %%xmm0   \n\t" /* filterCoeff */\
-        "movdqa              (%%"FF_REG_S", %%"FF_REG_c", 2), %%xmm2 \n\t" /* srcData */\
-        "movdqa            16(%%"FF_REG_S", %%"FF_REG_c", 2), %%xmm5 \n\t" /* srcData */\
-        "add                                $16, %%"FF_REG_d"        \n\t"\
-        "mov                        (%%"FF_REG_d"), %%"FF_REG_S"     \n\t"\
-        "test                         %%"FF_REG_S", %%"FF_REG_S"     \n\t"\
-        "pmulhw                           %%xmm0, %%xmm2      \n\t"\
-        "pmulhw                           %%xmm0, %%xmm5      \n\t"\
-        "paddw                            %%xmm2, %%xmm3      \n\t"\
-        "paddw                            %%xmm5, %%xmm4      \n\t"\
-        " jnz                                1b             \n\t"\
-        "psraw                               $3, %%xmm3      \n\t"\
-        "psraw                               $3, %%xmm4      \n\t"\
-        "packuswb                         %%xmm4, %%xmm3      \n\t"\
-        "movntdq                          %%xmm3, (%1, %%"FF_REG_c") \n\t"\
-        "add                         $16, %%"FF_REG_c"        \n\t"\
-        "cmp                          %2, %%"FF_REG_c"        \n\t"\
-        "movdqa                   %%xmm7, %%xmm3            \n\t" \
-        "movdqa                   %%xmm7, %%xmm4            \n\t" \
-        "mov                                 %0, %%"FF_REG_d"        \n\t"\
-        "mov                        (%%"FF_REG_d"), %%"FF_REG_S"     \n\t"\
-        "jb                                  1b             \n\t"
-
-    if (offset) {
-        __asm__ volatile(
-            "movq          %5, %%xmm3  \n\t"
-            "movdqa    %%xmm3, %%xmm4  \n\t"
-            "psrlq        $24, %%xmm3  \n\t"
-            "psllq        $40, %%xmm4  \n\t"
-            "por       %%xmm4, %%xmm3  \n\t"
-            MAIN_FUNCTION
-              :: "g" (filter),
-              "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset),
-              "m"(filterSize), "m"(((uint64_t *) dither)[0])
-              : XMM_CLOBBERS("%xmm0" , "%xmm1" , "%xmm2" , "%xmm3" , "%xmm4" , "%xmm5" , "%xmm7" ,)
-                "%"FF_REG_d, "%"FF_REG_S, "%"FF_REG_c
-              );
-    } else {
-        __asm__ volatile(
-            "movq          %5, %%xmm3   \n\t"
-            MAIN_FUNCTION
-              :: "g" (filter),
-              "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset),
-              "m"(filterSize), "m"(((uint64_t *) dither)[0])
-              : XMM_CLOBBERS("%xmm0" , "%xmm1" , "%xmm2" , "%xmm3" , "%xmm4" , "%xmm5" , "%xmm7" ,)
-                "%"FF_REG_d, "%"FF_REG_S, "%"FF_REG_c
-              );
-    }
+#define YUV2YUVX_FUNC_MMX(opt, step)  \
+void ff_yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, int srcOffset, \
+                           uint8_t *dest, int dstW,  \
+                           const uint8_t *dither, int offset); \
+static void yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, \
+                           const int16_t **src, uint8_t *dest, int dstW, \
+                           const uint8_t *dither, int offset) \
+{ \
+    ff_yuv2yuvX_ ##opt(filter, filterSize - 1, 0, dest - offset, dstW + offset, dither, offset); \
+    return; \
 }
+
+#define YUV2YUVX_FUNC(opt, step)  \
+void ff_yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, int srcOffset, \
+                           uint8_t *dest, int dstW,  \
+                           const uint8_t *dither, int offset); \
+static void yuv2yuvX_ ##opt(const int16_t *filter, int filterSize, \
+                           const int16_t **src, uint8_t *dest, int dstW, \
+                           const uint8_t *dither, int offset) \
+{ \
+    int remainder = (dstW % step); \
+    int pixelsProcessed = dstW - remainder; \
+    if(((uintptr_t)dest) & 15){ \
+        yuv2yuvX_mmx(filter, filterSize, src, dest, dstW, dither, offset); \
+        return; \
+    } \
+    ff_yuv2yuvX_ ##opt(filter, filterSize - 1, 0, dest - offset, pixelsProcessed + offset, dither, offset); \
+    if(remainder > 0){ \
+      ff_yuv2yuvX_mmx(filter, filterSize - 1, pixelsProcessed, dest - offset, pixelsProcessed + remainder + offset, dither, offset); \
+    } \
+    return; \
+}
+
+YUV2YUVX_FUNC_MMX(mmx, 16)
+YUV2YUVX_FUNC_MMX(mmxext, 16)
+YUV2YUVX_FUNC(sse3, 32)
+YUV2YUVX_FUNC(avx2, 64)
+
 #endif
 
 #endif /* HAVE_INLINE_ASM */
@@ -403,9 +376,14 @@ av_cold void ff_sws_init_swscale_x86(SwsContext *c)
 #if HAVE_MMXEXT_INLINE
     if (INLINE_MMXEXT(cpu_flags))
         sws_init_swscale_mmxext(c);
-    if (cpu_flags & AV_CPU_FLAG_SSE3){
-        if(c->use_mmx_vfilter && !(c->flags & SWS_ACCURATE_RND))
+    if (cpu_flags & AV_CPU_FLAG_AVX2){
+        if(c->use_mmx_vfilter && !(c->flags & SWS_ACCURATE_RND)){
+            c->yuv2planeX = yuv2yuvX_avx2;
+        }
+    } else if (cpu_flags & AV_CPU_FLAG_SSE3){
+        if(c->use_mmx_vfilter && !(c->flags & SWS_ACCURATE_RND)){
             c->yuv2planeX = yuv2yuvX_sse3;
+        }
     }
 #endif
 
diff --git a/libswscale/x86/swscale_template.c b/libswscale/x86/swscale_template.c
index 823056c2ea..cb33af97e4 100644
--- a/libswscale/x86/swscale_template.c
+++ b/libswscale/x86/swscale_template.c
@@ -38,88 +38,6 @@
 #endif
 #define MOVNTQ(a,b)  REAL_MOVNTQ(a,b)
 
-#if !COMPILE_TEMPLATE_MMXEXT
-static av_always_inline void
-dither_8to16(const uint8_t *srcDither, int rot)
-{
-    if (rot) {
-        __asm__ volatile("pxor      %%mm0, %%mm0\n\t"
-                         "movq       (%0), %%mm3\n\t"
-                         "movq      %%mm3, %%mm4\n\t"
-                         "psrlq       $24, %%mm3\n\t"
-                         "psllq       $40, %%mm4\n\t"
-                         "por       %%mm4, %%mm3\n\t"
-                         "movq      %%mm3, %%mm4\n\t"
-                         "punpcklbw %%mm0, %%mm3\n\t"
-                         "punpckhbw %%mm0, %%mm4\n\t"
-                         :: "r"(srcDither)
-                         );
-    } else {
-        __asm__ volatile("pxor      %%mm0, %%mm0\n\t"
-                         "movq       (%0), %%mm3\n\t"
-                         "movq      %%mm3, %%mm4\n\t"
-                         "punpcklbw %%mm0, %%mm3\n\t"
-                         "punpckhbw %%mm0, %%mm4\n\t"
-                         :: "r"(srcDither)
-                         );
-    }
-}
-#endif
-
-static void RENAME(yuv2yuvX)(const int16_t *filter, int filterSize,
-                           const int16_t **src, uint8_t *dest, int dstW,
-                           const uint8_t *dither, int offset)
-{
-    dither_8to16(dither, offset);
-    filterSize--;
-    __asm__ volatile(
-        "movd %0, %%mm1\n\t"
-        "punpcklwd %%mm1, %%mm1\n\t"
-        "punpckldq %%mm1, %%mm1\n\t"
-        "psllw        $3, %%mm1\n\t"
-        "paddw     %%mm1, %%mm3\n\t"
-        "paddw     %%mm1, %%mm4\n\t"
-        "psraw        $4, %%mm3\n\t"
-        "psraw        $4, %%mm4\n\t"
-        ::"m"(filterSize)
-     );
-
-    __asm__ volatile(\
-        "movq    %%mm3, %%mm6\n\t"
-        "movq    %%mm4, %%mm7\n\t"
-        "movl %3, %%ecx\n\t"
-        "mov                                 %0, %%"FF_REG_d"       \n\t"\
-        "mov                        (%%"FF_REG_d"), %%"FF_REG_S"    \n\t"\
-        ".p2align                             4                     \n\t" /* FIXME Unroll? */\
-        "1:                                                         \n\t"\
-        "movq                      8(%%"FF_REG_d"), %%mm0           \n\t" /* filterCoeff */\
-        "movq                (%%"FF_REG_S", %%"FF_REG_c", 2), %%mm2 \n\t" /* srcData */\
-        "movq               8(%%"FF_REG_S", %%"FF_REG_c", 2), %%mm5 \n\t" /* srcData */\
-        "add                                $16, %%"FF_REG_d"       \n\t"\
-        "mov                        (%%"FF_REG_d"), %%"FF_REG_S"    \n\t"\
-        "test                         %%"FF_REG_S", %%"FF_REG_S"    \n\t"\
-        "pmulhw                           %%mm0, %%mm2      \n\t"\
-        "pmulhw                           %%mm0, %%mm5      \n\t"\
-        "paddw                            %%mm2, %%mm3      \n\t"\
-        "paddw                            %%mm5, %%mm4      \n\t"\
-        " jnz                                1b             \n\t"\
-        "psraw                               $3, %%mm3      \n\t"\
-        "psraw                               $3, %%mm4      \n\t"\
-        "packuswb                         %%mm4, %%mm3      \n\t"
-        MOVNTQ2 "                         %%mm3, (%1, %%"FF_REG_c")\n\t"
-        "add                          $8, %%"FF_REG_c"      \n\t"\
-        "cmp                          %2, %%"FF_REG_c"      \n\t"\
-        "movq    %%mm6, %%mm3\n\t"
-        "movq    %%mm7, %%mm4\n\t"
-        "mov                                 %0, %%"FF_REG_d"     \n\t"\
-        "mov                        (%%"FF_REG_d"), %%"FF_REG_S"  \n\t"\
-        "jb                                  1b                   \n\t"\
-        :: "g" (filter),
-           "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset)
-        : "%"FF_REG_d, "%"FF_REG_S, "%"FF_REG_c
-    );
-}
-
 #define YSCALEYUV2PACKEDX_UV \
     __asm__ volatile(\
         "xor                %%"FF_REG_a", %%"FF_REG_a"  \n\t"\
diff --git a/libswscale/x86/yuv2yuvX.asm b/libswscale/x86/yuv2yuvX.asm
new file mode 100644
index 0000000000..b8874cde8c
--- /dev/null
+++ b/libswscale/x86/yuv2yuvX.asm
@@ -0,0 +1,136 @@
+;******************************************************************************
+;* x86-optimized yuv2yuvX
+;* Copyright 2020 Google LLC
+;* Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
+;*
+;* This file is part of FFmpeg.
+;*
+;* FFmpeg is free software; you can redistribute it and/or
+;* modify it under the terms of the GNU Lesser General Public
+;* License as published by the Free Software Foundation; either
+;* version 2.1 of the License, or (at your option) any later version.
+;*
+;* FFmpeg is distributed in the hope that it will be useful,
+;* but WITHOUT ANY WARRANTY; without even the implied warranty of
+;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+;* Lesser General Public License for more details.
+;*
+;* You should have received a copy of the GNU Lesser General Public
+;* License along with FFmpeg; if not, write to the Free Software
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+;******************************************************************************
+
+%include "libavutil/x86/x86util.asm"
+
+SECTION .text
+
+;-----------------------------------------------------------------------------
+; yuv2yuvX
+;
+; void ff_yuv2yuvX_<opt>(const int16_t *filter, int filterSize,
+;                        int srcOffset, uint8_t *dest, int dstW,
+;                        const uint8_t *dither, int offset);
+;
+;-----------------------------------------------------------------------------
+
+%macro YUV2YUVX_FUNC 0
+cglobal yuv2yuvX, 7, 7, 8, filter, filterSize, src, dest, dstW, dither, offset
+%if notcpuflag(sse3)
+%define movr mova
+%else
+%define movr movdqu
+%endif
+%if ARCH_X86_64
+    movsxd               dstWq, dstWd
+    movsxd               offsetq, offsetd
+    movsxd               srcq, srcd
+%endif ; x86-64
+%if cpuflag(avx2)
+    vpbroadcastq         m3, [ditherq]
+%elif cpuflag(sse3)
+    movq                 xmm3, [ditherq]
+%else
+    mova                 m3, [ditherq]
+%endif ; avx2
+    cmp                  offsetd, 0
+    jz                   .offset
+
+    ; offset != 0 path.
+    psrlq                m5, m3, $18
+    psllq                m3, m3, $28
+    por                  m3, m3, m5
+
+.offset:
+    add offsetq, srcq
+%if cpuflag(avx2)
+    movd                 xmm1, filterSized
+    vpbroadcastw         m1, xmm1
+%elif cpuflag(sse3)
+    movd                 xmm1, filterSized
+    pshuflw              m1, m1, q0000
+    punpcklqdq           m1, m1
+%else
+    movd m1, filterSized
+    punpcklwd m1, m1
+    punpckldq m1, m1
+%endif ; avx2
+    pxor                 m0, m0, m0
+    mov                  filterSizeq, filterq
+    mov                  srcq, [filterSizeq]
+    punpcklbw            m3, m0
+    psllw                m1, m1, 3
+    paddw                m3, m3, m1
+    psraw                m7, m3, 4
+.outerloop:
+    mova                 m4, m7
+    mova                 m3, m7
+    mova                 m6, m7
+    mova                 m1, m7
+.loop:
+%if cpuflag(avx2)
+    vpbroadcastq         m0, [filterSizeq + 8]
+%elif cpuflag(sse3)
+    movddup              m0, [filterSizeq + 8]
+%else
+    mova                 m0, [filterSizeq + 8]
+%endif
+    pmulhw               m2, m0, [srcq + offsetq * 2]
+    pmulhw               m5, m0, [srcq + offsetq * 2 + mmsize]
+    paddw                m3, m3, m2
+    paddw                m4, m4, m5
+    pmulhw               m2, m0, [srcq + offsetq * 2 + 2 * mmsize]
+    pmulhw               m5, m0, [srcq + offsetq * 2 + 3 * mmsize]
+    paddw                m6, m6, m2
+    paddw                m1, m1, m5
+    add                  filterSizeq, $10
+    mov                  srcq, [filterSizeq]
+    test                 srcq, srcq
+    jnz                  .loop
+    psraw                m3, m3, 3
+    psraw                m4, m4, 3
+    psraw                m6, m6, 3
+    psraw                m1, m1, 3
+    packuswb             m3, m3, m4
+    packuswb             m6, m6, m1
+    mov                  srcq, [filterq]
+%if cpuflag(avx2)
+    vpermq               m3, m3, 216
+    vpermq               m6, m6, 216
+%endif
+    movr                 [destq + offsetq], m3
+    movr                 [destq + offsetq + mmsize], m6
+    add                  offsetq, mmsize * 2
+    mov                  filterSizeq, filterq
+    cmp                  offsetq, dstWq
+    jb                  .outerloop
+    REP_RET
+%endmacro
+
+INIT_MMX mmx
+YUV2YUVX_FUNC
+INIT_MMX mmxext
+YUV2YUVX_FUNC
+INIT_XMM sse3
+YUV2YUVX_FUNC
+INIT_YMM avx2
+YUV2YUVX_FUNC