add MMX version of vp6_filter_diag

original patch by Sebastien Lucas  sebastien.lucas _at_ gmail _dot_ com

Originally committed as revision 17194 to svn://svn.ffmpeg.org/ffmpeg/trunk

3 files changed, 143 insertions, 0 deletions

diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 4b18873ed3..4d0503df9a 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -31,6 +31,7 @@
 #include "mmx.h"
 #include "vp3dsp_mmx.h"
 #include "vp3dsp_sse2.h"
+#include "vp6dsp_mmx.h"
 #include "idct_xvid.h"
 
 //#undef NDEBUG
@@ -2688,6 +2689,10 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
         c->h264_idct_add8      = ff_h264_idct_add8_mmx;
         c->h264_idct_add16intra= ff_h264_idct_add16intra_mmx;
 
+        if (CONFIG_VP6_DECODER) {
+            c->vp6_filter_diag4 = ff_vp6_filter_diag4_mmx;
+        }
+
         if (mm_flags & FF_MM_MMXEXT) {
             c->prefetch = prefetch_mmx2;
 
diff --git a/libavcodec/x86/vp6dsp_mmx.c b/libavcodec/x86/vp6dsp_mmx.c
new file mode 100644
index 0000000000..39e40d568e
--- /dev/null
+++ b/libavcodec/x86/vp6dsp_mmx.c
@@ -0,0 +1,108 @@
+/**
+ * @file libavcodec/x86/vp6dsp_mmx.c
+ * MMX-optimized functions for the VP6 decoder
+ *
+ * Copyright (C) 2009  Sebastien Lucas <sebastien.lucas@gmail.com>
+ *
+ * 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_cpu.h"
+#include "libavcodec/dsputil.h"
+#include "dsputil_mmx.h"
+#include "vp6dsp_mmx.h"
+
+
+#define DIAG4_MMX(in1,in2,in3,in4)                                            \
+    "movq  "#in1"(%0), %%mm0             \n\t"                                \
+    "movq  "#in2"(%0), %%mm1             \n\t"                                \
+    "movq       %%mm0, %%mm3             \n\t"                                \
+    "movq       %%mm1, %%mm4             \n\t"                                \
+    "punpcklbw  %%mm7, %%mm0             \n\t"                                \
+    "punpcklbw  %%mm7, %%mm1             \n\t"                                \
+    "punpckhbw  %%mm7, %%mm3             \n\t"                                \
+    "punpckhbw  %%mm7, %%mm4             \n\t"                                \
+    "pmullw     0(%2), %%mm0             \n\t" /* src[x-8 ] * biweight [0] */ \
+    "pmullw     8(%2), %%mm1             \n\t" /* src[x   ] * biweight [1] */ \
+    "pmullw     0(%2), %%mm3             \n\t" /* src[x-8 ] * biweight [0] */ \
+    "pmullw     8(%2), %%mm4             \n\t" /* src[x   ] * biweight [1] */ \
+    "paddw      %%mm1, %%mm0             \n\t"                                \
+    "paddw      %%mm4, %%mm3             \n\t"                                \
+    "movq  "#in3"(%0), %%mm1             \n\t"                                \
+    "movq  "#in4"(%0), %%mm2             \n\t"                                \
+    "movq       %%mm1, %%mm4             \n\t"                                \
+    "movq       %%mm2, %%mm5             \n\t"                                \
+    "punpcklbw  %%mm7, %%mm1             \n\t"                                \
+    "punpcklbw  %%mm7, %%mm2             \n\t"                                \
+    "punpckhbw  %%mm7, %%mm4             \n\t"                                \
+    "punpckhbw  %%mm7, %%mm5             \n\t"                                \
+    "pmullw    16(%2), %%mm1             \n\t" /* src[x+8 ] * biweight [2] */ \
+    "pmullw    24(%2), %%mm2             \n\t" /* src[x+16] * biweight [3] */ \
+    "pmullw    16(%2), %%mm4             \n\t" /* src[x+8 ] * biweight [2] */ \
+    "pmullw    24(%2), %%mm5             \n\t" /* src[x+16] * biweight [3] */ \
+    "paddw      %%mm2, %%mm1             \n\t"                                \
+    "paddw      %%mm5, %%mm4             \n\t"                                \
+    "paddsw     %%mm1, %%mm0             \n\t"                                \
+    "paddsw     %%mm4, %%mm3             \n\t"                                \
+    "paddsw     %%mm6, %%mm0             \n\t" /* Add 64 */                   \
+    "paddsw     %%mm6, %%mm3             \n\t" /* Add 64 */                   \
+    "psraw         $7, %%mm0             \n\t"                                \
+    "psraw         $7, %%mm3             \n\t"                                \
+    "packuswb   %%mm3, %%mm0             \n\t"                                \
+    "movq       %%mm0,  (%1)             \n\t"
+
+void ff_vp6_filter_diag4_mmx(uint8_t *dst, uint8_t *src, int stride,
+                             const int16_t *h_weights, const int16_t *v_weights)
+{
+    uint8_t tmp[8*11], *t = tmp;
+    int16_t weights[4*4];
+    int i;
+    src -= stride;
+
+    for (i=0; i<4*4; i++)
+        weights[i] = h_weights[i>>2];
+
+    __asm__ volatile(
+    "pxor %%mm7, %%mm7                   \n\t"
+    "movq "MANGLE(ff_pw_64)", %%mm6      \n\t"
+    "1:                                  \n\t"
+    DIAG4_MMX(-1,0,1,2)
+    "add  $8, %1                         \n\t"
+    "add  %3, %0                         \n\t"
+    "decl %4                             \n\t"
+    "jnz 1b                              \n\t"
+    : "+r"(src), "+r"(t)
+    : "r"(weights), "r"((x86_reg)stride), "r"(11)
+    : "memory");
+
+    t = tmp + 8;
+    for (i=0; i<4*4; i++)
+        weights[i] = v_weights[i>>2];
+
+    __asm__ volatile(
+    "pxor %%mm7, %%mm7                   \n\t"
+    "movq "MANGLE(ff_pw_64)", %%mm6      \n\t"
+    "1:                                  \n\t"
+    DIAG4_MMX(-8,0,8,16)
+    "add  $8, %0                         \n\t"
+    "add  %3, %1                         \n\t"
+    "decl %4                             \n\t"
+    "jnz 1b                              \n\t"
+    : "+r"(t), "+r"(dst)
+    : "r"(weights), "r"((x86_reg)stride), "r"(8)
+    : "memory");
+}
diff --git a/libavcodec/x86/vp6dsp_mmx.h b/libavcodec/x86/vp6dsp_mmx.h
new file mode 100644
index 0000000000..743bc4361d
--- /dev/null
+++ b/libavcodec/x86/vp6dsp_mmx.h
@@ -0,0 +1,30 @@
+/*
+ * vp6dsp MMX function declarations
+ * Copyright (c) 2009  Sebastien Lucas <sebastien.lucas@gmail.com>
+ *
+ * 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
+ */
+
+#ifndef AVCODEC_X86_VP6DSP_MMX_H
+#define AVCODEC_X86_VP6DSP_MMX_H
+
+#include <stdint.h>
+
+void ff_vp6_filter_diag4_mmx(uint8_t *dst, uint8_t *src, int stride,
+                             const int16_t *h_weights,const int16_t *v_weights);
+
+#endif /* AVCODEC_X86_VP6DSP_MMX_H */