]> git.sesse.net Git - ffmpeg/blobdiff - libavcodec/arm/vp8dsp_neon.S
projects / ffmpeg / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
imgconvert: cosmetics: Reshuffle defines to reduce ifdeffery
[ffmpeg] / libavcodec / arm / vp8dsp_neon.S
diff --git a/libavcodec/arm/vp8dsp_neon.S b/libavcodec/arm/vp8dsp_neon.S
index 92b1ef4eb7a3ba9455311ff0bbed1f003204ba05..544332c3c27e2aca2fa798f03d16da9016ec15e2 100644 (file)
--- a/libavcodec/arm/vp8dsp_neon.S
+++ b/libavcodec/arm/vp8dsp_neon.S
@@ -1,4 +1,4 @@
-/**
+/*
  * VP8 NEON optimisations
  *
  * Copyright (c) 2010 Rob Clark <rob@ti.com>
@@ -21,7 +21,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "asm.S"
+#include "libavutil/arm/asm.S"
+#include "neon.S"
 
 function ff_vp8_luma_dc_wht_neon, export=1
         vld1.16         {q0-q1},  [r1,:128]
@@ -76,18 +77,6 @@ function ff_vp8_luma_dc_wht_neon, export=1
         bx              lr
 endfunc
 
-function ff_vp8_luma_dc_wht_dc_neon, export=1
-        ldrsh           r2,  [r1]
-        mov             r3,  #0
-        add             r2,  r2,  #3
-        strh            r3,  [r1]
-        asr             r2,  r2,  #3
-    .rept 16
-        strh            r2,  [r0], #32
-    .endr
-        bx              lr
-endfunc
-
 function ff_vp8_idct_add_neon, export=1
         vld1.16         {q0-q1},  [r1,:128]
         movw            r3,  #20091
@@ -454,23 +443,6 @@ endfunc
     .endif
 .endm
 
-.macro transpose8x16matrix
-        vtrn.32         q0,   q4
-        vtrn.32         q1,   q5
-        vtrn.32         q2,   q6
-        vtrn.32         q3,   q7
-
-        vtrn.16         q0,   q2
-        vtrn.16         q1,   q3
-        vtrn.16         q4,   q6
-        vtrn.16         q5,   q7
-
-        vtrn.8          q0,   q1
-        vtrn.8          q2,   q3
-        vtrn.8          q4,   q5
-        vtrn.8          q6,   q7
-.endm
-
 .macro  vp8_v_loop_filter16 name, inner=0, simple=0
 function ff_vp8_v_loop_filter16\name\()_neon, export=1
         vpush           {q4-q7}
@@ -605,7 +577,7 @@ function ff_vp8_h_loop_filter16\name\()_neon, export=1
         vld1.8          {d13},    [r0], r1
         vld1.8          {d15},    [r0], r1
 
-        transpose8x16matrix
+        transpose_8x8   q0,  q1,  q2,  q3,  q4,  q5,  q6,  q7
 
         vdup.8          q14, r2                 @ flim_E
     .if !\simple
@@ -616,7 +588,7 @@ function ff_vp8_h_loop_filter16\name\()_neon, export=1
 
         sub             r0,  r0,  r1, lsl #4    @ backup 16 rows
 
-        transpose8x16matrix
+        transpose_8x8   q0,  q1,  q2,  q3,  q4,  q5,  q6,  q7
 
         @ Store pixels:
         vst1.8          {d0},     [r0],     r1
@@ -670,7 +642,7 @@ function ff_vp8_h_loop_filter8uv\name\()_neon, export=1
         vld1.8          {d14},    [r0], r2
         vld1.8          {d15},    [r1], r2
 
-        transpose8x16matrix
+        transpose_8x8   q0,  q1,  q2,  q3,  q4,  q5,  q6,  q7
 
         vdup.8          q14, r3                 @ flim_E
         vdup.8          q15, r12                @ flim_I
@@ -681,7 +653,7 @@ function ff_vp8_h_loop_filter8uv\name\()_neon, export=1
         sub             r0,  r0,  r2, lsl #3    @ backup u 8 rows
         sub             r1,  r1,  r2, lsl #3    @ backup v 8 rows
 
-        transpose8x16matrix
+        transpose_8x8   q0,  q1,  q2,  q3,  q4,  q5,  q6,  q7
 
         @ Store pixels:
         vst1.8          {d0},     [r0], r2
@@ -741,23 +713,6 @@ function ff_put_vp8_pixels8_neon, export=1
         bx              lr
 endfunc
 
-function ff_put_vp8_pixels4_neon, export=1
-        ldr             r12, [sp, #0]           @ h
-        push            {r4-r6,lr}
-1:
-        subs            r12, r12, #4
-        ldr             r4,       [r2], r3
-        ldr             r5,       [r2], r3
-        ldr             r6,       [r2], r3
-        ldr             lr,       [r2], r3
-        str             r4,       [r0], r1
-        str             r5,       [r0], r1
-        str             r6,       [r0], r1
-        str             lr,       [r0], r1
-        bgt             1b
-        pop             {r4-r6,pc}
-endfunc
-
 /* 4/6-tap 8th-pel MC */
 
 .macro  vp8_epel8_h6    d,   a,   b
@@ -1621,18 +1576,18 @@ endconst
 /* Bilinear MC */
 
 function ff_put_vp8_bilin16_h_neon, export=1
-        ldr             r3,  [sp, #4]           @ mx
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #4]           @ mx
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
         ldr             r12, [sp]               @ h
 1:
         subs            r12, r12, #2
-        vld1.8          {d2-d4},  [r2], r1
+        vld1.8          {d2-d4},  [r2], r3
         vext.8          q2,  q1,  q2,  #1
         vmull.u8        q8,  d2,  d1
         vmlal.u8        q8,  d4,  d0
-        vld1.8          {d18-d20},[r2], r1
+        vld1.8          {d18-d20},[r2], r3
         vmull.u8        q3,  d3,  d1
         vmlal.u8        q3,  d5,  d0
         vext.8          q10, q9,  q10, #1
@@ -1652,20 +1607,20 @@ function ff_put_vp8_bilin16_h_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin16_v_neon, export=1
-        ldr             r3,  [sp, #8]           @ my
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #8]           @ my
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
         ldr             r12, [sp]               @ h
-        vld1.8          {q1},     [r2], r1
+        vld1.8          {q1},     [r2], r3
 1:
         subs            r12, r12, #2
-        vld1.8          {q2},     [r2], r1
+        vld1.8          {q2},     [r2], r3
         vmull.u8        q3,  d2,  d1
         vmlal.u8        q3,  d4,  d0
         vmull.u8        q8,  d3,  d1
         vmlal.u8        q8,  d5,  d0
-        vld1.8          {q1},     [r2], r1
+        vld1.8          {q1},     [r2], r3
         vmull.u8        q9,  d4,  d1
         vmlal.u8        q9,  d2,  d0
         vmull.u8        q10, d5,  d1
@@ -1682,17 +1637,17 @@ function ff_put_vp8_bilin16_v_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin16_hv_neon, export=1
-        ldr             r3,  [sp, #4]           @ mx
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #4]           @ mx
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
-        ldr             r3,  [sp, #8]           @ my
-        rsb             r12, r3,  #8
-        vdup.8          d2,  r3
+        ldr             r12, [sp, #8]           @ my
+        vdup.8          d2,  r12
+        rsb             r12, r12, #8
         vdup.8          d3,  r12
         ldr             r12, [sp]               @ h
 
-        vld1.8          {d4-d6},  [r2], r1
+        vld1.8          {d4-d6},  [r2], r3
         vext.8          q3,  q2,  q3,  #1
         vmull.u8        q8,  d4,  d1
         vmlal.u8        q8,  d6,  d0
@@ -1702,11 +1657,11 @@ function ff_put_vp8_bilin16_hv_neon, export=1
         vrshrn.u16      d5,  q9,  #3
 1:
         subs            r12, r12, #2
-        vld1.8          {d18-d20},[r2], r1
+        vld1.8          {d18-d20},[r2], r3
         vext.8          q10, q9,  q10, #1
         vmull.u8        q11, d18, d1
         vmlal.u8        q11, d20, d0
-        vld1.8          {d26-d28},[r2], r1
+        vld1.8          {d26-d28},[r2], r3
         vmull.u8        q12, d19, d1
         vmlal.u8        q12, d21, d0
         vext.8          q14, q13, q14, #1
@@ -1738,18 +1693,18 @@ function ff_put_vp8_bilin16_hv_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin8_h_neon, export=1
-        ldr             r3,  [sp, #4]           @ mx
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #4]           @ mx
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
         ldr             r12, [sp]               @ h
 1:
         subs            r12, r12, #2
-        vld1.8          {q1},     [r2], r1
+        vld1.8          {q1},     [r2], r3
         vext.8          d3,  d2,  d3,  #1
         vmull.u8        q2,  d2,  d1
         vmlal.u8        q2,  d3,  d0
-        vld1.8          {q3},     [r2], r1
+        vld1.8          {q3},     [r2], r3
         vext.8          d7,  d6,  d7,  #1
         vmull.u8        q8,  d6,  d1
         vmlal.u8        q8,  d7,  d0
@@ -1763,18 +1718,18 @@ function ff_put_vp8_bilin8_h_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin8_v_neon, export=1
-        ldr             r3,  [sp, #8]           @ my
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #8]           @ my
+        vdup.8          d0,  r12
+        rsb             r12, r12,  #8
         vdup.8          d1,  r12
         ldr             r12, [sp]               @ h
-        vld1.8          {d2},     [r2], r1
+        vld1.8          {d2},     [r2], r3
 1:
         subs            r12, r12, #2
-        vld1.8          {d3},     [r2], r1
+        vld1.8          {d3},     [r2], r3
         vmull.u8        q2,  d2,  d1
         vmlal.u8        q2,  d3,  d0
-        vld1.8          {d2},     [r2], r1
+        vld1.8          {d2},     [r2], r3
         vmull.u8        q3,  d3,  d1
         vmlal.u8        q3,  d2,  d0
         vrshrn.u16      d4,  q2,  #3
@@ -1787,28 +1742,28 @@ function ff_put_vp8_bilin8_v_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin8_hv_neon, export=1
-        ldr             r3,  [sp, #4]           @ mx
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #4]           @ mx
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
-        ldr             r3,  [sp, #8]           @ my
-        rsb             r12, r3,  #8
-        vdup.8          d2,  r3
+        ldr             r12, [sp, #8]           @ my
+        vdup.8          d2,  r12
+        rsb             r12, r12, #8
         vdup.8          d3,  r12
         ldr             r12, [sp]               @ h
 
-        vld1.8          {q2},     [r2], r1
+        vld1.8          {q2},     [r2], r3
         vext.8          d5,  d4,  d5,  #1
         vmull.u8        q9,  d4,  d1
         vmlal.u8        q9,  d5,  d0
         vrshrn.u16      d22, q9,  #3
 1:
         subs            r12, r12, #2
-        vld1.8          {q3},     [r2], r1
+        vld1.8          {q3},     [r2], r3
         vext.8          d7,  d6,  d7,  #1
         vmull.u8        q8,  d6,  d1
         vmlal.u8        q8,  d7,  d0
-        vld1.8          {q2},     [r2], r1
+        vld1.8          {q2},     [r2], r3
         vext.8          d5,  d4,  d5,  #1
         vmull.u8        q9,  d4,  d1
         vmlal.u8        q9,  d5,  d0
@@ -1828,16 +1783,16 @@ function ff_put_vp8_bilin8_hv_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin4_h_neon, export=1
-        ldr             r3,  [sp, #4]           @ mx
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #4]           @ mx
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
         ldr             r12, [sp]               @ h
 1:
         subs            r12, r12, #2
-        vld1.8          {d2},     [r2], r1
+        vld1.8          {d2},     [r2], r3
         vext.8          d3,  d2,  d3,  #1
-        vld1.8          {d6},     [r2], r1
+        vld1.8          {d6},     [r2], r3
         vext.8          d7,  d6,  d7,  #1
         vtrn.32         q1,  q3
         vmull.u8        q2,  d2,  d1
@@ -1851,16 +1806,16 @@ function ff_put_vp8_bilin4_h_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin4_v_neon, export=1
-        ldr             r3,  [sp, #8]           @ my
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #8]           @ my
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
         ldr             r12, [sp]               @ h
-        vld1.32         {d2[]},   [r2], r1
+        vld1.32         {d2[]},   [r2], r3
 1:
         vld1.32         {d3[]},   [r2]
-        vld1.32         {d2[1]},  [r2], r1
-        vld1.32         {d3[1]},  [r2], r1
+        vld1.32         {d2[1]},  [r2], r3
+        vld1.32         {d3[1]},  [r2], r3
         vmull.u8        q2,  d2,  d1
         vmlal.u8        q2,  d3,  d0
         vtrn.32         d3,  d2
@@ -1874,26 +1829,26 @@ function ff_put_vp8_bilin4_v_neon, export=1
 endfunc
 
 function ff_put_vp8_bilin4_hv_neon, export=1
-        ldr             r3,  [sp, #4]           @ mx
-        rsb             r12, r3,  #8
-        vdup.8          d0,  r3
+        ldr             r12, [sp, #4]           @ mx
+        vdup.8          d0,  r12
+        rsb             r12, r12, #8
         vdup.8          d1,  r12
-        ldr             r3,  [sp, #8]           @ my
-        rsb             r12, r3,  #8
-        vdup.8          d2,  r3
+        ldr             r12, [sp, #8]           @ my
+        vdup.8          d2,  r12
+        rsb             r12, r12, #8
         vdup.8          d3,  r12
         ldr             r12, [sp]               @ h
 
-        vld1.8          {d4},     [r2], r1
+        vld1.8          {d4},     [r2], r3
         vext.8          d5,  d4,  d4,  #1
         vmull.u8        q9,  d4,  d1
         vmlal.u8        q9,  d5,  d0
         vrshrn.u16      d22, q9,  #3
 1:
         subs            r12, r12, #2
-        vld1.8          {d6},     [r2], r1
+        vld1.8          {d6},     [r2], r3
         vext.8          d7,  d6,  d6,  #1
-        vld1.8          {d4},     [r2], r1
+        vld1.8          {d4},     [r2], r3
         vext.8          d5,  d4,  d4,  #1
         vtrn.32         q3,  q2
         vmull.u8        q8,  d6,  d1
Unnamed repository; edit this file 'description' to name the repository.