Skip to content

F
ffmpeg.wasm-core
Commit e91339cd authored by Michael Niedermayer's avatar Michael Niedermayer
Browse files
Options

Merge commit '566b7a20' 

* commit '566b7a20':
  x86: float dsp: butterflies_float SSE

Conflicts:
	libavutil/x86/float_dsp.asm
Merged-by: 's avatarMichael Niedermayer <michaelni@gmx.at>
parents 550defe1 566b7a20
Hide whitespace changes
Inline Side-by-side
Showing with 17 additions and 15 deletions
libavutil/x86/float_dsp.asm
View file @ e91339cd
...@@ -268,21 +268,23 @@ cglobal scalarproduct_float, 3,3,2, v1, v2, offset ...@@ -268,21 +268,23 @@ cglobal scalarproduct_float, 3,3,2, v1, v2, offset
;----------------------------------------------------------------------------- ;-----------------------------------------------------------------------------
INIT_XMM sse INIT_XMM sse
cglobal butterflies_float, 3,3,3, src0, src1, len cglobal butterflies_float, 3,3,3, src0, src1, len
movsxdifnidn lenq, lend %if ARCH_X86_64
test lenq, lenq movsxd lenq, lend
jz .end %endif
shl lenq, 2 test lenq, lenq
add src0q, lenq jz .end
add src1q, lenq shl lenq, 2
neg lenq add src0q, lenq
add src1q, lenq
neg lenq
.loop: .loop:
mova m0, [src0q + lenq] mova m0, [src0q + lenq]
mova m1, [src1q + lenq] mova m1, [src1q + lenq]
subps m2, m0, m1 subps m2, m0, m1
addps m0, m0, m1 addps m0, m0, m1
mova [src1q + lenq], m2 mova [src1q + lenq], m2
mova [src0q + lenq], m0 mova [src0q + lenq], m0
add lenq, mmsize add lenq, mmsize
jl .loop jl .loop
.end: .end:
REP_RET REP_RET
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment