ARM: ARMv6 optimised add_pixels_clamped()
authorMåns Rullgård <mans@mansr.com>
Tue, 6 Oct 2009 21:55:35 +0000 (21:55 +0000)
committerMåns Rullgård <mans@mansr.com>
Tue, 6 Oct 2009 21:55:35 +0000 (21:55 +0000)
Originally committed as revision 20184 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavcodec/Makefile
libavcodec/arm/dsputil_armv6.S [new file with mode: 0644]
libavcodec/arm/dsputil_init_armv6.c

index 5666494..02e9465 100644 (file)
@@ -493,6 +493,7 @@ OBJS-$(HAVE_ARMV5TE)                   += arm/dsputil_init_armv5te.o    \
                                           arm/simple_idct_armv5te.o     \
 
 OBJS-$(HAVE_ARMV6)                     += arm/dsputil_init_armv6.o      \
+                                          arm/dsputil_armv6.o           \
                                           arm/simple_idct_armv6.o       \
 
 OBJS-$(HAVE_ARMVFP)                    += arm/dsputil_vfp.o             \
diff --git a/libavcodec/arm/dsputil_armv6.S b/libavcodec/arm/dsputil_armv6.S
new file mode 100644 (file)
index 0000000..c76d408
--- /dev/null
@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2009 Mans Rullgard <mans@mansr.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 "asm.S"
+
+        .text
+
+function ff_add_pixels_clamped_armv6, export=1
+        push            {r4-r8,lr}
+        mov             r3,  #8
+1:
+        ldm             r0!, {r4,r5,r12,lr}
+        ldrd            r6,  r7,  [r1]
+        pkhbt           r8,  r4,  r5,  lsl #16
+        pkhtb           r5,  r5,  r4,  asr #16
+        pkhbt           r4,  r12, lr,  lsl #16
+        pkhtb           lr,  lr,  r12, asr #16
+        pld             [r1, r2]
+        uxtab16         r8,  r8,  r6
+        uxtab16         r5,  r5,  r6,  ror #8
+        uxtab16         r4,  r4,  r7
+        uxtab16         lr,  lr,  r7,  ror #8
+        usat16          r8,  #8,  r8
+        usat16          r5,  #8,  r5
+        usat16          r4,  #8,  r4
+        usat16          lr,  #8,  lr
+        orr             r6,  r8,  r5,  lsl #8
+        orr             r7,  r4,  lr,  lsl #8
+        subs            r3,  r3,  #1
+        strd            r6,  r7,  [r1],  r2
+        bgt             1b
+        pop             {r4-r8,pc}
+.endfunc
index 6b03a67..76bde43 100644 (file)
@@ -25,6 +25,10 @@ void ff_simple_idct_armv6(DCTELEM *data);
 void ff_simple_idct_put_armv6(uint8_t *dest, int line_size, DCTELEM *data);
 void ff_simple_idct_add_armv6(uint8_t *dest, int line_size, DCTELEM *data);
 
+void ff_add_pixels_clamped_armv6(const DCTELEM *block,
+                                 uint8_t *restrict pixels,
+                                 int line_size);
+
 void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx)
 {
     if (!avctx->lowres && (avctx->idct_algo == FF_IDCT_AUTO ||
@@ -34,4 +38,6 @@ void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx)
         c->idct                  = ff_simple_idct_armv6;
         c->idct_permutation_type = FF_LIBMPEG2_IDCT_PERM;
     }
+
+    c->add_pixels_clamped = ff_add_pixels_clamped_armv6;
 }