wrapper/libavc
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

arm: Use neg instead of rsb with a #0 immediate

Browse source 
Change-Id: I52cc96b15a553e3c757d2a627c1e2201fc832f24
This commit is contained in:
Martin Storsjo 2016-10-05 11:10:10 +03:00
parent 6958350a65
commit 37edf81bad
4 changed files with 6 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

2
common/arm/ih264_intra_pred_luma_16x16_a9q.s
View file

@ -413,7 +413,7 @@ scrlbl1:
add r7, r0, r4, lsl #3 add r7, r0, r4, lsl #3
sub r0, r7, r4, lsl #1 sub r0, r7, r4, lsl #1
rsb lr, r4, #0x0 neg lr, r4
vpadd.s16 d0, d0, d1 vpadd.s16 d0, d0, d1

2
common/arm/ih264_padding_neon.s
View file

@ -88,7 +88,7 @@ ih264_pad_top_a9q:
stmfd sp!, {r4-r11, lr} @stack stores the values of the arguments stmfd sp!, {r4-r11, lr} @stack stores the values of the arguments
sub r5, r0, r1 sub r5, r0, r1
rsb r6, r1, #0 neg r6, r1
loop_neon_memcpy_mul_16: loop_neon_memcpy_mul_16:
@ Load 16 bytes @ Load 16 bytes

4
common/arm/ih264_weighted_bi_pred_a9q.s
View file

@ -144,7 +144,7 @@ ih264_weighted_bi_pred_luma_a9q:
ldr r4, [sp, #40] @Load src_strd2 in r4 ldr r4, [sp, #40] @Load src_strd2 in r4
ldr r5, [sp, #44] @Load dst_strd in r5 ldr r5, [sp, #44] @Load dst_strd in r5
sxtb r9, r9 @sign-extend 8-bit ofst1 to 32-bit sxtb r9, r9 @sign-extend 8-bit ofst1 to 32-bit
rsb r10, r6, #0 @r10 = -(log_wd + 1) neg r10, r6 @r10 = -(log_wd + 1)
ldr r11, [sp, #68] @Load ht in r11 ldr r11, [sp, #68] @Load ht in r11
ldr r12, [sp, #72] @Load wd in r12 ldr r12, [sp, #72] @Load wd in r12
vdup.16 q0, r10 @Q0 = -(log_wd + 1) (32-bit) vdup.16 q0, r10 @Q0 = -(log_wd + 1) (32-bit)
@ -456,7 +456,7 @@ ih264_weighted_bi_pred_chroma_a9q:
ldr r9, [sp, #60] @Load ofst1 in r9 ldr r9, [sp, #60] @Load ofst1 in r9
ldr r10, [sp, #64] @Load ofst2 in r10 ldr r10, [sp, #64] @Load ofst2 in r10
rsb r12, r6, #0 @r12 = -(log_wd + 1) neg r12, r6 @r12 = -(log_wd + 1)
ldr r4, [sp, #40] @Load src_strd2 in r4 ldr r4, [sp, #40] @Load src_strd2 in r4
ldr r5, [sp, #44] @Load dst_strd in r5 ldr r5, [sp, #44] @Load dst_strd in r5
vdup.16 q0, r12 @Q0 = -(log_wd + 1) (16-bit) vdup.16 q0, r12 @Q0 = -(log_wd + 1) (16-bit)

4
common/arm/ih264_weighted_pred_a9q.s
View file

@ -122,7 +122,7 @@ ih264_weighted_pred_luma_a9q:
vpush {d8-d15} vpush {d8-d15}
vdup.16 d2, r5 @D2 = wt (16-bit) vdup.16 d2, r5 @D2 = wt (16-bit)
rsb r9, r4, #0 @r9 = -log_wd neg r9, r4 @r9 = -log_wd
vdup.8 d3, r6 @D3 = ofst (8-bit) vdup.8 d3, r6 @D3 = ofst (8-bit)
cmp r8, #16 @check if wd is 16 cmp r8, #16 @check if wd is 16
vdup.16 q0, r9 @Q0 = -log_wd (16-bit) vdup.16 q0, r9 @Q0 = -log_wd (16-bit)
@ -349,7 +349,7 @@ ih264_weighted_pred_chroma_a9q:
ldr r6, [sp, #36] @Load ofst = {ofst_u (8-bit), ofst_v (8-bit)} ldr r6, [sp, #36] @Load ofst = {ofst_u (8-bit), ofst_v (8-bit)}
ldr r8, [sp, #44] @Load wd ldr r8, [sp, #44] @Load wd
rsb r9, r4, #0 @r9 = -log_wd neg r9, r4 @r9 = -log_wd
vdup.32 q1, r5 @Q1 = {wt_u (16-bit), wt_v (16-bit)} vdup.32 q1, r5 @Q1 = {wt_u (16-bit), wt_v (16-bit)}
ldr r7, [sp, #40] @Load ht ldr r7, [sp, #40] @Load ht
vpush {d8-d15} vpush {d8-d15}