[FFmpeg-devel] [PATCH] ARM: NEON optimised put_pixels functions

Mans Rullgard mans
Mon Aug 25 05:06:34 CEST 2008


---
 libavcodec/Makefile                |    3 +
 libavcodec/armv4l/dsputil_arm.c    |    4 +
 libavcodec/armv4l/dsputil_neon.c   |   67 ++++++++++
 libavcodec/armv4l/dsputil_neon_s.S |  254 ++++++++++++++++++++++++++++++++++++
 4 files changed, 328 insertions(+), 0 deletions(-)
 create mode 100644 libavcodec/armv4l/dsputil_neon.c
 create mode 100644 libavcodec/armv4l/dsputil_neon_s.S

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 03a5df6..43f8324 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -431,7 +431,10 @@ ASM_OBJS-$(HAVE_ARMV5TE)               += armv4l/simple_idct_armv5te.o \
 
 ASM_OBJS-$(HAVE_ARMV6)                 += armv4l/simple_idct_armv6.o \
 
+OBJS-$(HAVE_NEON)                      += armv4l/dsputil_neon.o \
+
 ASM_OBJS-$(HAVE_NEON)                  += armv4l/simple_idct_neon.o \
+                                          armv4l/dsputil_neon_s.o   \
 
 OBJS-$(HAVE_VIS)                       += sparc/dsputil_vis.o \
                                           sparc/simple_idct_vis.o \
diff --git a/libavcodec/armv4l/dsputil_arm.c b/libavcodec/armv4l/dsputil_arm.c
index 2d0759a..9f5ca31 100644
--- a/libavcodec/armv4l/dsputil_arm.c
+++ b/libavcodec/armv4l/dsputil_arm.c
@@ -26,6 +26,7 @@
 
 extern void dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);
 extern void ff_float_init_arm_vfp(DSPContext* c, AVCodecContext *avctx);
+extern void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx);
 
 extern void j_rev_dct_ARM(DCTELEM *data);
 extern void simple_idct_ARM(DCTELEM *data);
@@ -317,4 +318,7 @@ void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx)
 #ifdef HAVE_ARMVFP
     ff_float_init_arm_vfp(c, avctx);
 #endif
+#ifdef HAVE_NEON
+    ff_dsputil_init_neon(c, avctx);
+#endif
 }
diff --git a/libavcodec/armv4l/dsputil_neon.c b/libavcodec/armv4l/dsputil_neon.c
new file mode 100644
index 0000000..8a10dde
--- /dev/null
+++ b/libavcodec/armv4l/dsputil_neon.c
@@ -0,0 +1,67 @@
+/*
+ * ARM NEON optimised DSP functions
+ * Copyright (c) 2008 Mans Rullgard <mans at 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 <stdint.h>
+
+#include "libavcodec/avcodec.h"
+#include "libavcodec/dsputil.h"
+
+void ff_put_pixels16_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+
+void ff_put_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
+void ff_put_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int);
+
+void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
+{
+    c->put_pixels_tab[0][0] = ff_put_pixels16_neon;
+    c->put_pixels_tab[0][1] = ff_put_pixels16_x2_neon;
+    c->put_pixels_tab[0][2] = ff_put_pixels16_y2_neon;
+    c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_neon;
+    c->put_pixels_tab[1][0] = ff_put_pixels8_neon;
+    c->put_pixels_tab[1][1] = ff_put_pixels8_x2_neon;
+    c->put_pixels_tab[1][2] = ff_put_pixels8_y2_neon;
+    c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_neon;
+
+    c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_neon;
+    c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_neon;
+    c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_neon;
+
+    c->put_h264_qpel_pixels_tab[0][0] = ff_put_h264_qpel16_mc00_neon;
+    c->put_h264_qpel_pixels_tab[1][0] = ff_put_h264_qpel8_mc00_neon;
+}
diff --git a/libavcodec/armv4l/dsputil_neon_s.S b/libavcodec/armv4l/dsputil_neon_s.S
new file mode 100644
index 0000000..b8e1520
--- /dev/null
+++ b/libavcodec/armv4l/dsputil_neon_s.S
@@ -0,0 +1,254 @@
+/*
+ * ARM NEON optimised DSP functions
+ * Copyright (c) 2008 Mans Rullgard <mans at 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
+ */
+
+        .fpu neon
+        .text
+
+        .macro put_pixels16
+        dmb
+1:      vld1.64         {d0, d1},  [r1], r2
+        vld1.64         {d2, d3},  [r1], r2
+        vld1.64         {d4, d5},  [r1], r2
+        vld1.64         {d6, d7},  [r1], r2
+        pld             [r1]
+        subs            r3,  r3,  #4
+        vst1.64         {d0, d1},  [r0,:128], r2
+        vst1.64         {d2, d3},  [r0,:128], r2
+        vst1.64         {d4, d5},  [r0,:128], r2
+        vst1.64         {d6, d7},  [r0,:128], r2
+        bne             1b
+        bx              lr
+        .endm
+
+        .macro put_pixels16_x2 vhadd=vrhadd.u8
+        dmb
+1:      vld1.64         {d0-d2},   [r1], r2
+        vld1.64         {d4-d6},   [r1], r2
+        pld             [r1]
+        subs            r3,  r3,  #2
+        vext.8          q1,  q0,  q1,  #1
+        vext.8          q3,  q2,  q3,  #1
+        \vhadd          q0,  q0,  q1
+        \vhadd          q2,  q2,  q3
+        vst1.64         {d0, d1},  [r0,:64], r2
+        vst1.64         {d4, d5},  [r0,:64], r2
+        bne             1b
+        bx              lr
+        .endm
+
+        .macro put_pixels16_y2 vhadd=vrhadd.u8
+        push            {lr}
+        add             ip,  r1,  r2
+        lsl             lr,  r2,  #1
+        vld1.64         {d0, d1},  [r1], lr
+        vld1.64         {d2, d3},  [ip], lr
+        dmb
+1:      subs            r3,  r3,  #2
+        \vhadd          q2,  q0,  q1
+        vld1.64         {d0, d1},  [r1],      lr
+        vst1.64         {d4, d5},  [r0,:128], r2
+        \vhadd          q2,  q0,  q1
+        vld1.64         {d2, d3},  [ip],      lr
+        vst1.64         {d4, d5},  [r0,:128], r2
+        bne             1b
+        pop             {pc}
+        .endm
+
+        .macro put_pixels16_xy2 vshrn=vrshrn.u16 no_rnd=0
+        push            {lr}
+        lsl             lr,  r2,  #1
+        add             ip,  r1,  r2
+        vld1.64         {d0-d2},   [r1], lr
+        vld1.64         {d4-d6},   [ip], lr
+        .if \no_rnd
+        vmov.i16        q13, #1
+        .endif
+        pld             [r1]
+        pld             [ip]
+        vext.8          q1,  q0,  q1,  #1
+        vext.8          q3,  q2,  q3,  #1
+        vaddl.u8        q8,  d0,  d2
+        vaddl.u8        q10, d1,  d3
+        vaddl.u8        q9,  d4,  d6
+        vaddl.u8        q11, d5,  d7
+        dmb
+1:      subs            r3,  r3,  #2
+        vld1.64         {d0-d2},   [r1], lr
+        vadd.u16        q12, q8,  q9
+        pld             [r1]
+        .if \no_rnd
+        vadd.u16        q12, q12, q13
+        .endif
+        vext.8          q15, q0,  q1,  #1
+        vadd.u16        q1 , q10, q11
+        \vshrn          d28, q12, #2
+        .if \no_rnd
+        vadd.u16        q1,  q1,  q13
+        .endif
+        \vshrn          d29, q1,  #2
+        vaddl.u8        q8,  d0,  d30
+        vld1.64         {d2-d4},   [ip], lr
+        vaddl.u8        q10, d1,  d31
+        vst1.64         {d28,d29}, [r0,:128], r2
+        vadd.u16        q12, q8,  q9
+        pld             [ip]
+        .if \no_rnd
+        vadd.u16        q12, q12, q13
+        .endif
+        vext.8          q2,  q1,  q2,  #1
+        vadd.u16        q0,  q10, q11
+        \vshrn          d30, q12, #2
+        .if \no_rnd
+        vadd.u16        q0,  q0,  q13
+        .endif
+        \vshrn          d31, q0,  #2
+        vaddl.u8        q9,  d2,  d4
+        vaddl.u8        q11, d3,  d5
+        vst1.64         {d30,d31}, [r0,:128], r2
+        bgt             1b
+        pop             {pc}
+        .endm
+
+        .macro put_pixels8
+        dmb
+1:      vld1.64         {d0}, [r1], r2
+        vld1.64         {d1}, [r1], r2
+        vld1.64         {d2}, [r1], r2
+        vld1.64         {d3}, [r1], r2
+        subs            r3,  r3,  #4
+        vst1.64         {d0}, [r0,:64], r2
+        vst1.64         {d1}, [r0,:64], r2
+        vst1.64         {d2}, [r0,:64], r2
+        vst1.64         {d3}, [r0,:64], r2
+        bne             1b
+        bx              lr
+        .endm
+
+        .macro put_pixels8_x2 vhadd=vrhadd.u8
+        dmb
+1:      vld1.64         {d0, d1},  [r1], r2
+        vld1.64         {d2, d3},  [r1], r2
+        pld             [r1]
+        subs            r3,  r3,  #2
+        vext.8          d1,  d0,  d1,  #1
+        vext.8          d3,  d2,  d3,  #1
+        vswp            d1,  d2
+        \vhadd          q0,  q0,  q1
+        vst1.64         {d0},      [r0,:64], r2
+        vst1.64         {d1},      [r0,:64], r2
+        bne             1b
+        bx              lr
+        .endm
+
+        .macro put_pixels8_y2 vhadd=vrhadd.u8
+        push            {lr}
+        add             ip,  r1,  r2
+        lsl             lr,  r2,  #1
+        vld1.64         {d0},      [r1], lr
+        vld1.64         {d1},      [ip], lr
+        dmb
+1:      subs            r3,  r3,  #2
+        \vhadd          d4,  d0,  d1
+        vld1.64         {d0},      [r1],     lr
+        vst1.64         {d4},      [r0,:64], r2
+        \vhadd          d4,  d0,  d1
+        vld1.64         {d1},      [ip],     lr
+        vst1.64         {d4},      [r0,:64], r2
+        bne             1b
+        pop             {pc}
+        .endm
+
+        .macro put_pixels8_xy2 vshrn=vrshrn.u16 no_rnd=0
+        push            {lr}
+        lsl             lr,  r2,  #1
+        add             ip,  r1,  r2
+        vld1.64         {d0, d1},  [r1], lr
+        vld1.64         {d2, d3},  [ip], lr
+	.if \no_rnd
+        vmov.i16        q11, #1
+        .endif
+        pld             [r1]
+        pld             [ip]
+        vext.8          d4,  d0,  d1,  #1
+        vext.8          d6,  d2,  d3,  #1
+        vaddl.u8        q8,  d0,  d4
+        vaddl.u8        q9,  d2,  d6
+        dmb
+1:      subs            r3,  r3,  #2
+        vld1.64         {d0, d1},  [r1], lr
+        pld             [r1]
+        vadd.u16        q10, q8,  q9
+        vext.8          d4,  d0,  d1,  #1
+	.if \no_rnd
+        vadd.u16        q10, q10, q11
+        .endif
+        vaddl.u8        q8,  d0,  d4
+        \vshrn          d5,  q10, #2
+        vld1.64         {d2, d3},  [ip], lr
+        vadd.u16        q10, q8,  q9
+        pld             [ip]
+	.if \no_rnd
+        vadd.u16        q10, q10, q11
+        .endif
+        vst1.64         {d5},      [r0,:64], r2
+        \vshrn          d7,  q10, #2
+        vext.8          d6,  d2,  d3,  #1
+        vaddl.u8        q9,  d2,  d6
+        vst1.64         {d7},      [r0,:64], r2
+        bgt             1b
+        pop             {pc}
+        .endm
+
+        .macro extern name
+        .global \name
+        .type   \name, %function
+        .func   \name
+\name:
+        .endm
+
+        .macro defun name suf rnd_op args:vararg
+        extern ff_\name\suf\()_neon
+        \name \rnd_op \args
+        .endfunc
+        .endm
+
+        .macro defun2 name args:vararg
+        defun \name
+        defun \name \args
+        .endm
+
+        extern ff_put_h264_qpel16_mc00_neon
+        mov   r3, #16
+        .endfunc
+
+        defun  put_pixels16
+        defun2 put_pixels16_x2,  _no_rnd, vhadd.u8
+        defun2 put_pixels16_y2,  _no_rnd, vhadd.u8
+        defun2 put_pixels16_xy2, _no_rnd, vshrn.u16, 1
+
+        extern ff_put_h264_qpel8_mc00_neon
+        mov   r3, #8
+        .endfunc
+
+        defun  put_pixels8
+        defun2 put_pixels8_x2,   _no_rnd, vhadd.u8
+        defun2 put_pixels8_y2,   _no_rnd, vhadd.u8
+        defun2 put_pixels8_xy2,  _no_rnd, vshrn.u16, 1
-- 
1.6.0





More information about the ffmpeg-devel mailing list