UltraSPARC VIS yuv2rgb

patch by Denes Balatoni: [dbalatoni interware hu]
original thread: [FFmpeg-devel] [PATCH] UltraSPARC VIS yuv2rgb third try
date: 07/17/2007 08:35 PM

Originally committed as revision 23825 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

3 files changed, 219 insertions, 0 deletions

diff --git a/libswscale/swscale_internal.h b/libswscale/swscale_internal.h
index baebdf2..c31a443 100644
--- a/libswscale/swscale_internal.h
+++ b/libswscale/swscale_internal.h
@@ -177,6 +177,9 @@ typedef struct SwsContext{
     uint32_t gmask        __attribute__((aligned(4)));
 #endif
 
+#ifdef HAVE_VIS
+    uint64_t sparc_coeffs[10] __attribute__((aligned(8)));
+#endif
 
 } SwsContext;
 //FIXME check init (where 0)
diff --git a/libswscale/yuv2rgb.c b/libswscale/yuv2rgb.c
index 5262c20..065c0e8 100644
--- a/libswscale/yuv2rgb.c
+++ b/libswscale/yuv2rgb.c
@@ -39,6 +39,10 @@
 #include "swscale.h"
 #include "swscale_internal.h"
 
+#ifdef HAVE_VIS
+#include "yuv2rgb_vis.c"
+#endif
+
 #ifdef HAVE_MLIB
 #include "yuv2rgb_mlib.c"
 #endif
@@ -630,6 +634,12 @@ SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
         }
     }
 #endif
+#ifdef HAVE_VIS
+    {
+        SwsFunc t= yuv2rgb_init_vis(c);
+        if (t) return t;
+    }
+#endif
 #ifdef HAVE_MLIB
     {
         SwsFunc t= yuv2rgb_init_mlib(c);
diff --git a/libswscale/yuv2rgb_vis.c b/libswscale/yuv2rgb_vis.c
new file mode 100644
index 0000000..b0b531c
--- /dev/null
+++ b/libswscale/yuv2rgb_vis.c
@@ -0,0 +1,206 @@
+/*
+ * VIS optimized software YUV to RGB converter
+ * Copyright (c) 2007 Denes Balatoni <dbalatoni@programozo.hu>
+ *
+ * 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 <inttypes.h>
+#include <stdlib.h>
+
+#include "swscale.h"
+
+#define YUV2RGB_INIT \
+    "wr %%g0, 0x10, %%gsr \n\t" \
+    "ldd [%5], %%f32      \n\t" \
+    "ldd [%5+8], %%f34    \n\t" \
+    "ldd [%5+16], %%f36   \n\t" \
+    "ldd [%5+24], %%f38   \n\t" \
+    "ldd [%5+32], %%f40   \n\t" \
+    "ldd [%5+40], %%f42   \n\t" \
+    "ldd [%5+48], %%f44   \n\t" \
+    "ldd [%5+56], %%f46   \n\t" \
+    "ldd [%5+64], %%f48   \n\t" \
+    "ldd [%5+72], %%f50   \n\t"
+
+#define YUV2RGB_KERNEL \
+    /* ^^^^ f0=Y f3=u f5=v */ \
+    "fmul8x16 %%f3, %%f48, %%f6   \n\t" \
+    "fmul8x16 %%f19, %%f48, %%f22 \n\t" \
+    "fmul8x16 %%f5, %%f44, %%f8   \n\t" \
+    "fmul8x16 %%f21, %%f44, %%f24 \n\t" \
+    "fmul8x16 %%f0, %%f42, %%f0   \n\t" \
+    "fmul8x16 %%f16, %%f42, %%f16 \n\t" \
+    "fmul8x16 %%f3, %%f50, %%f2   \n\t" \
+    "fmul8x16 %%f19, %%f50, %%f18 \n\t" \
+    "fmul8x16 %%f5, %%f46, %%f4   \n\t" \
+    "fmul8x16 %%f21, %%f46, %%f20 \n\t" \
+    \
+    "fpsub16 %%f6, %%f34, %%f6   \n\t" /* 1 */ \
+    "fpsub16 %%f22, %%f34, %%f22 \n\t" /* 1 */ \
+    "fpsub16 %%f8, %%f38, %%f8   \n\t" /* 3 */ \
+    "fpsub16 %%f24, %%f38, %%f24 \n\t" /* 3 */ \
+    "fpsub16 %%f0, %%f32, %%f0   \n\t" /* 0 */ \
+    "fpsub16 %%f16, %%f32, %%f16 \n\t" /* 0 */ \
+    "fpsub16 %%f2, %%f36, %%f2   \n\t" /* 2 */ \
+    "fpsub16 %%f18, %%f36, %%f18 \n\t" /* 2 */ \
+    "fpsub16 %%f4, %%f40, %%f4   \n\t" /* 4 */ \
+    "fpsub16 %%f20, %%f40, %%f20 \n\t" /* 4 */ \
+    \
+    "fpadd16 %%f0, %%f8, %%f8    \n\t" /* Gt */ \
+    "fpadd16 %%f16, %%f24, %%f24 \n\t" /* Gt */ \
+    "fpadd16 %%f0, %%f4, %%f4    \n\t" /* R */ \
+    "fpadd16 %%f16, %%f20, %%f20 \n\t" /* R */ \
+    "fpadd16 %%f0, %%f6, %%f6    \n\t" /* B */ \
+    "fpadd16 %%f16, %%f22, %%f22 \n\t" /* B */ \
+    "fpadd16 %%f8, %%f2, %%f2    \n\t" /* G */ \
+    "fpadd16 %%f24, %%f18, %%f18 \n\t" /* G */ \
+    \
+    "fpack16 %%f4, %%f4    \n\t" \
+    "fpack16 %%f20, %%f20  \n\t" \
+    "fpack16 %%f6, %%f6    \n\t" \
+    "fpack16 %%f22, %%f22  \n\t" \
+    "fpack16 %%f2, %%f2    \n\t" \
+    "fpack16 %%f18, %%f18  \n\t"
+
+
+
+static int vis_420P_ARGB32(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
+                           int srcSliceH, uint8_t* dst[], int dstStride[]){
+  int y, out1, out2, out3, out4, out5, out6;
+
+  for(y=0;y < srcSliceH;++y) {
+      asm volatile (
+          YUV2RGB_INIT
+          "wr %%g0, 0xd2, %%asi        \n\t" /* ASI_FL16_P */
+          "1:                          \n\t"
+          "ldda [%1] %%asi, %%f2       \n\t"
+          "ldda [%1+2] %%asi, %%f18    \n\t"
+          "ldda [%2] %%asi, %%f4       \n\t"
+          "ldda [%2+2] %%asi, %%f20    \n\t"
+          "ld [%0], %%f0               \n\t"
+          "ld [%0+4], %%f16            \n\t"
+          "fpmerge %%f3, %%f3, %%f2    \n\t"
+          "fpmerge %%f19, %%f19, %%f18 \n\t"
+          "fpmerge %%f5, %%f5, %%f4    \n\t"
+          "fpmerge %%f21, %%f21, %%f20 \n\t"
+          YUV2RGB_KERNEL
+          "fzero %%f0                  \n\t"
+          "fpmerge %%f4, %%f6, %%f8    \n\t"  // r,b,t1
+          "fpmerge %%f20, %%f22, %%f24 \n\t"  // r,b,t1
+          "fpmerge %%f0, %%f2, %%f10   \n\t"  // 0,g,t2
+          "fpmerge %%f0, %%f18, %%f26  \n\t"  // 0,g,t2
+          "fpmerge %%f10, %%f8, %%f4   \n\t"  // t2,t1,msb
+          "fpmerge %%f26, %%f24, %%f20 \n\t"  // t2,t1,msb
+          "fpmerge %%f11, %%f9, %%f6   \n\t"  // t2,t1,lsb
+          "fpmerge %%f27, %%f25, %%f22 \n\t"  // t2,t1,lsb
+          "std %%f4, [%3]              \n\t"
+          "std %%f20, [%3+16]          \n\t"
+          "std %%f6, [%3+8]            \n\t"
+          "std %%f22, [%3+24]          \n\t"
+
+          "add %0, 8, %0   \n\t"
+          "add %1, 4, %1   \n\t"
+          "add %2, 4, %2   \n\t"
+          "subcc %4, 8, %4 \n\t"
+          "bne 1b          \n\t"
+          "add %3, 32, %3  \n\t" //delay slot
+          : "=r" (out1), "=r" (out2), "=r" (out3), "=r" (out4), "=r" (out5), "=r" (out6)
+          : "0" (src[0]+(y+srcSliceY)*srcStride[0]), "1" (src[1]+((y+srcSliceY)>>1)*srcStride[1]),
+            "2" (src[2]+((y+srcSliceY)>>1)*srcStride[2]), "3" (dst[0]+(y+srcSliceY)*dstStride[0]),
+            "4" (c->dstW),
+            "5" (c->sparc_coeffs)
+      );
+  }
+
+  return srcSliceH;
+}
+
+static int vis_422P_ARGB32(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
+                           int srcSliceH, uint8_t* dst[], int dstStride[]){
+  int y, out1, out2, out3, out4, out5, out6;
+
+  for(y=0;y < srcSliceH;++y) {
+      asm volatile (
+          YUV2RGB_INIT
+          "wr %%g0, 0xd2, %%asi        \n\t" /* ASI_FL16_P */
+          "1:                          \n\t"
+          "ldda [%1] %%asi, %%f2       \n\t"
+          "ldda [%1+2] %%asi, %%f18    \n\t"
+          "ldda [%2] %%asi, %%f4       \n\t"
+          "ldda [%2+2] %%asi, %%f20    \n\t"
+          "ld [%0], %%f0               \n\t"
+          "ld [%0+4], %%f16            \n\t"
+          "fpmerge %%f3, %%f3, %%f2    \n\t"
+          "fpmerge %%f19, %%f19, %%f18 \n\t"
+          "fpmerge %%f5, %%f5, %%f4    \n\t"
+          "fpmerge %%f21, %%f21, %%f20 \n\t"
+          YUV2RGB_KERNEL
+          "fzero %%f0 \n\t"
+          "fpmerge %%f4, %%f6, %%f8    \n\t"  // r,b,t1
+          "fpmerge %%f20, %%f22, %%f24 \n\t"  // r,b,t1
+          "fpmerge %%f0, %%f2, %%f10   \n\t"  // 0,g,t2
+          "fpmerge %%f0, %%f18, %%f26  \n\t"  // 0,g,t2
+          "fpmerge %%f10, %%f8, %%f4   \n\t"  // t2,t1,msb
+          "fpmerge %%f26, %%f24, %%f20 \n\t"  // t2,t1,msb
+          "fpmerge %%f11, %%f9, %%f6   \n\t"  // t2,t1,lsb
+          "fpmerge %%f27, %%f25, %%f22 \n\t"  // t2,t1,lsb
+          "std %%f4, [%3]              \n\t"
+          "std %%f20, [%3+16]          \n\t"
+          "std %%f6, [%3+8]            \n\t"
+          "std %%f22, [%3+24]          \n\t"
+
+          "add %0, 8, %0   \n\t"
+          "add %1, 4, %1   \n\t"
+          "add %2, 4, %2   \n\t"
+          "subcc %4, 8, %4 \n\t"
+          "bne 1b          \n\t"
+          "add %3, 32, %3  \n\t" //delay slot
+          : "=r" (out1), "=r" (out2), "=r" (out3), "=r" (out4), "=r" (out5), "=r" (out6)
+          : "0" (src[0]+(y+srcSliceY)*srcStride[0]), "1" (src[1]+(y+srcSliceY)*srcStride[1]),
+            "2" (src[2]+(y+srcSliceY)*srcStride[2]), "3" (dst[0]+(y+srcSliceY)*dstStride[0]),
+            "4" (c->dstW),
+            "5" (c->sparc_coeffs)
+      );
+  }
+
+  return srcSliceH;
+}
+
+SwsFunc yuv2rgb_init_vis(SwsContext *c) {
+    c->sparc_coeffs[5]=c->yCoeff;
+    c->sparc_coeffs[6]=c->vgCoeff;
+    c->sparc_coeffs[7]=c->vrCoeff;
+    c->sparc_coeffs[8]=c->ubCoeff;
+    c->sparc_coeffs[9]=c->ugCoeff;
+
+    c->sparc_coeffs[0]=(((int16_t)c->yOffset*(int16_t)c->yCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
+    c->sparc_coeffs[1]=(((int16_t)c->uOffset*(int16_t)c->ubCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
+    c->sparc_coeffs[2]=(((int16_t)c->uOffset*(int16_t)c->ugCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
+    c->sparc_coeffs[3]=(((int16_t)c->vOffset*(int16_t)c->vgCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
+    c->sparc_coeffs[4]=(((int16_t)c->vOffset*(int16_t)c->vrCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
+
+    if (c->dstFormat == PIX_FMT_RGB32 && c->srcFormat == PIX_FMT_YUV422P && (c->dstW & 7)==0) {
+        av_log(c, AV_LOG_INFO, "SPARC VIS accelerated YUV422P -> RGB32\n");
+        return vis_422P_ARGB32;
+    }
+    else if (c->dstFormat == PIX_FMT_RGB32 && c->srcFormat == PIX_FMT_YUV420P && (c->dstW & 7)==0) {
+        av_log(c, AV_LOG_INFO, "SPARC VIS accelerated YUV420P -> RGB32\n");
+        return vis_420P_ARGB32;
+    }
+    return NULL;
+}