generic-library/openssl
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

rc4-x86_64.pl: "Westmere" optimization.

Browse Source
This commit is contained in:
Andy Polyakov
2010-05-13 21:01:24 +00:00
parent 1aa8a6297c
commit 629fd3aa91
1 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
crypto/rc4/asm/rc4-x86_64.pl
View File

@@ -58,6 +58,10 @@
# fit for Core2 and therefore the code was modified to skip cloop8 on # fit for Core2 and therefore the code was modified to skip cloop8 on
# this CPU. # this CPU.
# Intel Westmere was observed to perform suboptimally. Adding yet
# another movzb to cloop1 improved performance by almost 50%! Core2
# performance is improved too, but nominally...
$flavour = shift; $flavour = shift;
$output = shift; $output = shift;
if ($flavour =~ /\./) { $output = $flavour; undef $flavour; } if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
@@ -229,6 +233,7 @@ $code.=<<___;
.align 16 .align 16
.Lcloop1: .Lcloop1:
add $TX[0]#b,$YY#b add $TX[0]#b,$YY#b
movzb $YY#b,$YY#d
movzb ($dat,$YY),$TY#d movzb ($dat,$YY),$TY#d
movb $TX[0]#b,($dat,$YY) movb $TX[0]#b,($dat,$YY)
movb $TY#b,($dat,$XX[0]) movb $TY#b,($dat,$XX[0])