x86/v210dec: use named registers

Signed-off-by: James Almer <jamrial@gmail.com>
pull/310/head
James Almer 6 years ago
parent abf1aa87ab
commit b41d8ab2e6
  1. 32
      libavcodec/x86/v210.asm

@ -39,12 +39,12 @@ SECTION .text
%macro v210_planar_unpack 1
; v210_planar_unpack(const uint32_t *src, uint16_t *y, uint16_t *u, uint16_t *v, int width)
cglobal v210_planar_unpack_%1, 5, 5, 6 + 2 * cpuflag(avx2)
movsxdifnidn r4, r4d
lea r1, [r1+2*r4]
add r2, r4
add r3, r4
neg r4
cglobal v210_planar_unpack_%1, 5, 5, 6 + 2 * cpuflag(avx2), src, y, u, v, w
movsxdifnidn wq, wd
lea yq, [yq+2*wq]
add uq, wq
add vq, wq
neg wq
VBROADCASTI128 m3, [v210_mult]
@ -60,9 +60,9 @@ cglobal v210_planar_unpack_%1, 5, 5, 6 + 2 * cpuflag(avx2)
.loop:
%ifidn %1, unaligned
movu m0, [r0] ; yB v5 yA u5 y9 v4 y8 u4 y7 v3 y6 u3 y5 v2 y4 u2 y3 v1 y2 u1 y1 v0 y0 u0
movu m0, [srcq] ; yB v5 yA u5 y9 v4 y8 u4 y7 v3 y6 u3 y5 v2 y4 u2 y3 v1 y2 u1 y1 v0 y0 u0
%else
mova m0, [r0]
mova m0, [srcq]
%endif
pmullw m1, m0, m3
@ -74,29 +74,29 @@ cglobal v210_planar_unpack_%1, 5, 5, 6 + 2 * cpuflag(avx2)
vpblendd m2, m1, m0, 0x55 ; yB yA 00 y9 y8 y7 00 y6 y5 y4 00 y3 y2 y1 00 y0
pshufb m2, m4 ; 00 00 yB yA y9 y8 y7 y6 00 00 y5 y4 y3 y2 y1 y0
vpermd m2, m6, m2 ; 00 00 00 00 yB yA y9 y8 y7 y6 y5 y4 y3 y2 y1 y0
movu [r1+2*r4], m2
movu [yq+2*wq], m2
vpblendd m1, m1, m0, 0xaa ; 00 v5 u5 v4 00 u4 v3 u3 00 v2 u2 v1 00 u1 v0 u0
pshufb m1, m5 ; 00 v5 v4 v3 00 u5 u4 u3 00 v2 v1 v0 00 u2 u1 u0
vpermq m1, m1, 0xd8 ; 00 v5 v4 v3 00 v2 v1 v0 00 u5 u4 u3 00 u2 u1 u0
pshufb m1, m7 ; 00 00 v5 v4 v3 v2 v1 v0 00 00 u5 u4 u3 u2 u1 u0
movu [r2+r4], xm1
vextracti128 [r3+r4], m1, 1
movu [uq+wq], xm1
vextracti128 [vq+wq], m1, 1
%else
shufps m2, m1, m0, 0x8d ; 00 y9 00 y6 yB yA y8 y7 00 y3 00 y0 y5 y4 y2 y1
pshufb m2, m4 ; 00 00 yB yA y9 y8 y7 y6 00 00 y5 y4 y3 y2 y1 y0
movu [r1+2*r4], m2
movu [yq+2*wq], m2
shufps m1, m0, 0xd8 ; 00 v5 00 u4 u5 v4 v3 u3 00 v2 00 u1 u2 v1 v0 u0
pshufb m1, m5 ; 00 v5 v4 v3 00 u5 u4 u3 00 v2 v1 v0 00 u2 u1 u0
movq [r2+r4], m1
movhps [r3+r4], m1
movq [uq+wq], m1
movhps [vq+wq], m1
%endif
add r0, mmsize
add r4, (mmsize*3)/8
add srcq, mmsize
add wq, (mmsize*3)/8
jl .loop
REP_RET

Loading…
Cancel
Save