Commit 649b7a99 authored by Pascal Massimino's avatar Pascal Massimino Committed by James Almer

av_filter/x86/idet: use HADDD where appropriate

Signed-off-by: 's avatarJames Almer <jamrial@gmail.com>
parent c7d8dbad
...@@ -69,10 +69,8 @@ cglobal idet_filter_line, 4, 5, 0, a, b, c, width, index ...@@ -69,10 +69,8 @@ cglobal idet_filter_line, 4, 5, 0, a, b, c, width, index
CMP widthd, indexd CMP widthd, indexd
jg .loop jg .loop
mova m0, m_sum HADDD m_sum, m0
psrlq m_sum, 0x20 movd eax, m_sum
paddd m0, m_sum
movd eax, m0
RET RET
%endmacro %endmacro
...@@ -123,18 +121,7 @@ cglobal idet_filter_line_16bit, 4, 5, 8, a, b, c, width, index ...@@ -123,18 +121,7 @@ cglobal idet_filter_line_16bit, 4, 5, 8, a, b, c, width, index
paddd m_sum, m5 paddd m_sum, m5
jl .loop_16bit jl .loop_16bit
mova m2, m_sum HADDD m_sum, m2
%if mmsize == 16
psrldq m2, 4
paddd m_sum, m2
psrldq m2, 4
paddd m_sum, m2
psrldq m2, 4
paddd m_sum, m2
%else
psrlq m2, 32
paddd m_sum, m2
%endif
movd eax, m_sum movd eax, m_sum
RET RET
%endmacro %endmacro
......
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