highperfocused/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

x86/aacdec: use HADDPS macro

Browse Source 
Signed-off-by: James Almer <jamrial@gmail.com>
This commit is contained in:
James Almer 2016-06-08 14:18:18 -03:00
parent 99b899483e
commit 82dbfccaf0
2 changed files with 7 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
libavcodec/x86/aacpsdsp.asm
View File

@ -38,17 +38,7 @@ cglobal ps_add_squares, 3, 3, %1, dst, src, n
movaps m1, [srcq+mmsize] movaps m1, [srcq+mmsize]
mulps m0, m0 mulps m0, m0
mulps m1, m1 mulps m1, m1
%if cpuflag(sse3) HADDPS m0, m1, m2
haddps m0, m1
%else
movaps m3, m0
movaps m4, m1
shufps m3, m3, q0301
shufps m4, m4, q0301
addps m0, m3
addps m1, m4
shufps m0, m1, q2020
%endif
addps m0, [dstq] addps m0, [dstq]
movaps [dstq], m0 movaps [dstq], m0
add dstq, mmsize add dstq, mmsize
@ -59,9 +49,9 @@ cglobal ps_add_squares, 3, 3, %1, dst, src, n
%endmacro %endmacro
INIT_XMM sse INIT_XMM sse
PS_ADD_SQUARES 3 PS_ADD_SQUARES 2
INIT_XMM sse3 INIT_XMM sse3
PS_ADD_SQUARES 5 PS_ADD_SQUARES 3
;******************************************************************* ;*******************************************************************
;void ff_ps_mul_pair_single_sse(float (*dst)[2], float (*src0)[2], ;void ff_ps_mul_pair_single_sse(float (*dst)[2], float (*src0)[2],

24
libavcodec/x86/sbrdsp.asm
View File

@ -526,31 +526,15 @@ align 16
xorps m2, [ps_mask3] xorps m2, [ps_mask3]
xorps m5, [ps_mask3] xorps m5, [ps_mask3]
xorps m6, [ps_mask3] xorps m6, [ps_mask3]
HADDPS m2, m5, m3
HADDPS m7, m6, m4
%if cpuflag(sse3) %if cpuflag(sse3)
movshdup m0, m1 movshdup m0, m1
haddps m2, m5
haddps m7, m6
addss m1, m0
%else %else
movaps m3, m2 movss m0, m1
movaps m0, m5
movaps m4, m6
shufps m3, m3, q0301
shufps m0, m0, q0301
shufps m4, m4, q0301
addps m2, m3
addps m5, m0
addps m6, m4
movss m0, m7
movss m3, m1
shufps m7, m7, q0001
shufps m1, m1, q0001 shufps m1, m1, q0001
addss m7, m0
addss m1, m3
shufps m2, m5, q2020
shufps m7, m6, q2020
%endif %endif
addss m1, m0
movaps [phiq ], m2 movaps [phiq ], m2
movhps [phiq+0x18], m7 movhps [phiq+0x18], m7
movss [phiq+0x28], m7 movss [phiq+0x28], m7