diff options
| author | Andy Polyakov <appro@openssl.org> | 2004-05-31 12:06:27 +0000 |
|---|---|---|
| committer | Andy Polyakov <appro@openssl.org> | 2004-05-31 12:06:27 +0000 |
| commit | a2eb9688a4e63eb78e2068986d0cfacd501aa30e (patch) | |
| tree | ab658802e791944a95e9a3fbe5bf6fc5c807f6b8 /crypto/sha | |
| parent | af2bf07404f393a9b7425a943918830ae08413d3 (diff) | |
Kill unused macro and reimplement it for that single context it can
actually be used, namely x86* platforms [because they don't bomb on
unaligned access]. This resulted in 30-40% [depending on message
length] improvement for SHA-256 compiled with gcc and running on P4.
In the lack of assembler implementation I give the compiler all the
help it can possibly get:-)
Diffstat (limited to 'crypto/sha')
| -rw-r--r-- | crypto/sha/asm/sha512-sse2.pl | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/crypto/sha/asm/sha512-sse2.pl b/crypto/sha/asm/sha512-sse2.pl index 4235dbae82..797aedacd7 100644 --- a/crypto/sha/asm/sha512-sse2.pl +++ b/crypto/sha/asm/sha512-sse2.pl @@ -21,11 +21,11 @@ # Throughput performance in MBps (larger is better): # # 2.4GHz P4 1.4GHz AMD32 1.4GHz AMD64(*) -# SHA256/gcc(*) 39 42 59 +# SHA256/gcc(*) 54 43 59 # SHA512/gcc 17 23 92 # SHA512/sse2 54(**) 55(**) # SHA512/icc 26 28 -# SHA256/icc(*) 64 54 +# SHA256/icc(*) 65 54 # # (*) AMD64 and SHA256 numbers are presented mostly for amusement or # reference purposes. |