Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(14)

Issues Search
    My Issues | Starred     Open | Closed | All

Unified Diff: source/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm

Issue 7671004: Update libvpx snapshot to v0.9.7-p1 (Cayuga). (Closed) Base URL: svn://chrome-svn/chrome/trunk/deps/third_party/libvpx/
Patch Set: '' Created 9 years, 4 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View side-by-side diff with in-line comments
Download patch
« no previous file with comments | « source/libvpx/vp8/common/arm/armv6/filter_v6.asm ('k') | source/libvpx/vp8/common/arm/armv6/simpleloopfilter_v6.asm » ('j') | no next file with comments »
Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
Index: source/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm
===================================================================
--- source/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm (revision 96967)
+++ source/libvpx/vp8/common/arm/armv6/loopfilter_v6.asm (working copy)
@@ -53,14 +53,11 @@
;r0 unsigned char *src_ptr,
;r1 int src_pixel_step,
-;r2 const char *flimit,
+;r2 const char *blimit,
;r3 const char *limit,
;stack const char *thresh,
;stack int count
-;Note: All 16 elements in flimit are equal. So, in the code, only one load is needed
-;for flimit. Same way applies to limit and thresh.
-
;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
|vp8_loop_filter_horizontal_edge_armv6| PROC
;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
@@ -72,14 +69,18 @@
sub sp, sp, #16 ; create temp buffer
ldr r9, [src], pstep ; p3
- ldr r4, [r2], #4 ; flimit
+ ldrb r4, [r2] ; blimit
ldr r10, [src], pstep ; p2
- ldr r2, [r3], #4 ; limit
+ ldrb r2, [r3] ; limit
ldr r11, [src], pstep ; p1
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r6], #4 ; thresh
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r6] ; thresh
+ orr r2, r2, r2, lsl #8
mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
|Hnext8|
; vp8_filter_mask() function
@@ -253,12 +254,6 @@
subs count, count, #1
- ;pld [src]
- ;pld [src, pstep]
- ;pld [src, pstep, lsl #1]
- ;pld [src, pstep, lsl #2]
- ;pld [src, pstep, lsl #3]
-
ldrne r9, [src], pstep ; p3
ldrne r10, [src], pstep ; p2
ldrne r11, [src], pstep ; p1
@@ -281,14 +276,18 @@
sub sp, sp, #16 ; create temp buffer
ldr r9, [src], pstep ; p3
- ldr r4, [r2], #4 ; flimit
+ ldrb r4, [r2] ; blimit
ldr r10, [src], pstep ; p2
- ldr r2, [r3], #4 ; limit
+ ldrb r2, [r3] ; limit
ldr r11, [src], pstep ; p1
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r6], #4 ; thresh
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r6] ; thresh
+ orr r2, r2, r2, lsl #8
mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
|MBHnext8|
@@ -590,15 +589,19 @@
sub sp, sp, #16 ; create temp buffer
ldr r6, [src], pstep ; load source data
- ldr r4, [r2], #4 ; flimit
+ ldrb r4, [r2] ; blimit
ldr r7, [src], pstep
- ldr r2, [r3], #4 ; limit
+ ldrb r2, [r3] ; limit
ldr r8, [src], pstep
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r12], #4 ; thresh
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r12] ; thresh
+ orr r2, r2, r2, lsl #8
ldr lr, [src], pstep
mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
|Vnext8|
@@ -857,18 +860,26 @@
sub src, src, #4 ; move src pointer down by 4
ldr count, [sp, #40] ; count for 8-in-parallel
ldr r12, [sp, #36] ; load thresh address
+ pld [src, #23] ; preload for next block
sub sp, sp, #16 ; create temp buffer
ldr r6, [src], pstep ; load source data
- ldr r4, [r2], #4 ; flimit
+ ldrb r4, [r2] ; blimit
+ pld [src, #23]
ldr r7, [src], pstep
- ldr r2, [r3], #4 ; limit
+ ldrb r2, [r3] ; limit
+ pld [src, #23]
ldr r8, [src], pstep
- uadd8 r4, r4, r4 ; flimit * 2
- ldr r3, [r12], #4 ; thresh
+ orr r4, r4, r4, lsl #8
+ ldrb r3, [r12] ; thresh
+ orr r2, r2, r2, lsl #8
+ pld [src, #23]
ldr lr, [src], pstep
mov count, count, lsl #1 ; 4-in-parallel
- uadd8 r4, r4, r2 ; flimit * 2 + limit
+ orr r4, r4, r4, lsl #16
+ orr r3, r3, r3, lsl #8
+ orr r2, r2, r2, lsl #16
+ orr r3, r3, r3, lsl #16
|MBVnext8|
; vp8_filter_mask() function
@@ -908,6 +919,7 @@
str lr, [sp, #8]
ldr lr, [src], pstep
+
TRANSPOSE_MATRIX r6, r7, r8, lr, r9, r10, r11, r12
ldr lr, [sp, #8] ; load back (f)limit accumulator
@@ -956,6 +968,7 @@
beq mbvskip_filter ; skip filtering
+
;vp8_hevmask() function
;calculate high edge variance
@@ -1123,6 +1136,7 @@
smlabb r8, r6, lr, r7
smlatb r6, r6, lr, r7
smlabb r9, r10, lr, r7
+
smlatb r10, r10, lr, r7
ssat r8, #8, r8, asr #7
ssat r6, #8, r6, asr #7
@@ -1242,9 +1256,13 @@
sub src, src, #4
subs count, count, #1
+ pld [src, #23] ; preload for next block
ldrne r6, [src], pstep ; load source data
+ pld [src, #23]
ldrne r7, [src], pstep
+ pld [src, #23]
ldrne r8, [src], pstep
+ pld [src, #23]
ldrne lr, [src], pstep
bne MBVnext8
« no previous file with comments | « source/libvpx/vp8/common/arm/armv6/filter_v6.asm ('k') | source/libvpx/vp8/common/arm/armv6/simpleloopfilter_v6.asm » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698