diff options
| author | Martin Storsjö <martin@martin.st> | 2017-01-09 00:04:19 +0200 |
|---|---|---|
| committer | Martin Storsjö <martin@martin.st> | 2017-03-19 22:53:32 +0200 |
| commit | 21c89f3a26bb1331381b90e653277585447cfbb3 (patch) | |
| tree | e29745bb6f8296b5a7952721a48822ab645003b2 /libavcodec/aarch64 | |
| parent | 70317b25aa35c0907720e4d2b7686408588c07aa (diff) | |
| download | ffmpeg-21c89f3a26bb1331381b90e653277585447cfbb3.tar.gz | |
arm/aarch64: vp9: Fix vertical alignment
Align the second/third operands as they usually are.
Due to the wildly varying sizes of the written out operands
in aarch64 assembly, the column alignment is usually not as clear
as in arm assembly.
This is cherrypicked from libav commit
7995ebfad12002033c73feed422a1cfc62081e8f.
Signed-off-by: Martin Storsjö <martin@martin.st>
Diffstat (limited to 'libavcodec/aarch64')
| -rw-r--r-- | libavcodec/aarch64/vp9itxfm_neon.S | 36 |
1 files changed, 18 insertions, 18 deletions
diff --git a/libavcodec/aarch64/vp9itxfm_neon.S b/libavcodec/aarch64/vp9itxfm_neon.S index 3e5da0880c..b12890f0db 100644 --- a/libavcodec/aarch64/vp9itxfm_neon.S +++ b/libavcodec/aarch64/vp9itxfm_neon.S @@ -380,7 +380,7 @@ function ff_vp9_\txfm1\()_\txfm2\()_8x8_add_neon, export=1 .ifc \txfm1\()_\txfm2,idct_idct movrel x4, idct_coeffs .else - movrel x4, iadst8_coeffs + movrel x4, iadst8_coeffs ld1 {v1.8h}, [x4], #16 .endif ld1 {v0.8h}, [x4] @@ -480,23 +480,23 @@ itxfm_func8x8 iadst, iadst function idct16x16_dc_add_neon - movrel x4, idct_coeffs + movrel x4, idct_coeffs ld1 {v0.4h}, [x4] - movi v1.4h, #0 + movi v1.4h, #0 ld1 {v2.h}[0], [x2] - smull v2.4s, v2.4h, v0.h[0] - rshrn v2.4h, v2.4s, #14 - smull v2.4s, v2.4h, v0.h[0] - rshrn v2.4h, v2.4s, #14 + smull v2.4s, v2.4h, v0.h[0] + rshrn v2.4h, v2.4s, #14 + smull v2.4s, v2.4h, v0.h[0] + rshrn v2.4h, v2.4s, #14 dup v2.8h, v2.h[0] st1 {v1.h}[0], [x2] - srshr v2.8h, v2.8h, #6 + srshr v2.8h, v2.8h, #6 - mov x3, x0 - mov x4, #16 + mov x3, x0 + mov x4, #16 1: // Loop to add the constant from v2 into all 16x16 outputs subs x4, x4, #2 @@ -869,7 +869,7 @@ function ff_vp9_\txfm1\()_\txfm2\()_16x16_add_neon, export=1 .ifc \txfm1,idct ld1 {v0.8h,v1.8h}, [x10] .endif - mov x9, #32 + mov x9, #32 .ifc \txfm1\()_\txfm2,idct_idct cmp w3, #10 @@ -1046,10 +1046,10 @@ idct16_partial quarter idct16_partial half function idct32x32_dc_add_neon - movrel x4, idct_coeffs + movrel x4, idct_coeffs ld1 {v0.4h}, [x4] - movi v1.4h, #0 + movi v1.4h, #0 ld1 {v2.h}[0], [x2] smull v2.4s, v2.4h, v0.h[0] @@ -1059,10 +1059,10 @@ function idct32x32_dc_add_neon dup v2.8h, v2.h[0] st1 {v1.h}[0], [x2] - srshr v0.8h, v2.8h, #6 + srshr v0.8h, v2.8h, #6 - mov x3, x0 - mov x4, #32 + mov x3, x0 + mov x4, #32 1: // Loop to add the constant v0 into all 32x32 outputs subs x4, x4, #2 @@ -1230,7 +1230,7 @@ endfunc // x9 = double input stride function idct32_1d_8x32_pass1\suffix\()_neon mov x14, x30 - movi v2.8h, #0 + movi v2.8h, #0 // v16 = IN(0), v17 = IN(2) ... v31 = IN(30) .ifb \suffix @@ -1295,7 +1295,7 @@ function idct32_1d_8x32_pass1\suffix\()_neon .endif add x2, x2, #64 - movi v2.8h, #0 + movi v2.8h, #0 // v16 = IN(1), v17 = IN(3) ... v31 = IN(31) .ifb \suffix .irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 |