RepoMirrors/ffmpeg
1
0
Fork 0
mirror of https://git.ffmpeg.org/ffmpeg.git synced 2025-05-01 15:48:03 +00:00
Code Issues Projects Releases Wiki Activity

pixblockdsp: x86: Condense diff_pixels_* to a shared macro

Browse Source 
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Reviewed-by: James Almer <jamrial@gmail.com>
This commit is contained in:
Timothy Gu 2015-11-01 03:12:45 -08:00
parent 7cea3430a5
commit 4b80b895a9
1 changed files with 30 additions and 34 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
libavcodec/x86/pixblockdsp.asm
View File

@ -80,54 +80,50 @@ cglobal get_pixels, 3, 4, 5
mova [r0+0x70], m3 mova [r0+0x70], m3
RET RET
INIT_MMX mmx
; void ff_diff_pixels_mmx(int16_t *block, const uint8_t *s1, const uint8_t *s2, ; void ff_diff_pixels_mmx(int16_t *block, const uint8_t *s1, const uint8_t *s2,
; int stride); ; int stride);
cglobal diff_pixels, 4,5 %macro DIFF_PIXELS 0
movsxdifnidn r3, r3d
pxor m7, m7
add r0, 128
mov r4, -128
.loop:
mova m0, [r1]
mova m2, [r2]
mova m1, m0
mova m3, m2
punpcklbw m0, m7
punpckhbw m1, m7
punpcklbw m2, m7
punpckhbw m3, m7
psubw m0, m2
psubw m1, m3
mova [r0+r4+0], m0
mova [r0+r4+8], m1
add r1, r3
add r2, r3
add r4, 16
jne .loop
REP_RET
INIT_XMM sse2
cglobal diff_pixels, 4,5,5 cglobal diff_pixels, 4,5,5
movsxdifnidn r3, r3d movsxdifnidn r3, r3d
pxor m4, m4 pxor m4, m4
add r0, 128 add r0, 128
mov r4, -128 mov r4, -128
.loop: .loop:
movh m0, [r1] movq m0, [r1]
movh m2, [r2] movq m2, [r2]
movh m1, [r1+r3] %if mmsize == 8
movh m3, [r2+r3] movq m1, m0
movq m3, m2
punpcklbw m0, m4
punpckhbw m1, m4
punpcklbw m2, m4
punpckhbw m3, m4
%else
movq m1, [r1+r3]
movq m3, [r2+r3]
punpcklbw m0, m4 punpcklbw m0, m4
punpcklbw m1, m4 punpcklbw m1, m4
punpcklbw m2, m4 punpcklbw m2, m4
punpcklbw m3, m4 punpcklbw m3, m4
%endif
psubw m0, m2 psubw m0, m2
psubw m1, m3 psubw m1, m3
mova [r0+r4+0], m0 mova [r0+r4+0], m0
mova [r0+r4+16], m1 mova [r0+r4+mmsize], m1
%if mmsize == 8
add r1, r3
add r2, r3
%else
lea r1, [r1+r3*2] lea r1, [r1+r3*2]
lea r2, [r2+r3*2] lea r2, [r2+r3*2]
add r4, 32 %endif
add r4, 2 * mmsize
jne .loop jne .loop
RET RET
%endmacro
INIT_MMX mmx
DIFF_PIXELS
INIT_XMM sse2
DIFF_PIXELS