Commentary updates.

This commit is contained in:
Andy Polyakov 2007-09-13 07:27:10 +00:00
parent 1fa29843fa
commit 399f94bfb4
5 changed files with 9 additions and 5 deletions

View File

@ -18,6 +18,7 @@
# -m64 -m32 # -m64 -m32
# ---------------------------------- # ----------------------------------
# PPC970,gcc-4.0.0 +76% +59% # PPC970,gcc-4.0.0 +76% +59%
# Power6,xlc-7 +68% +33%
$output = shift; $output = shift;

View File

@ -27,7 +27,7 @@
# gcc 3.4 32-bit asm cycles/byte # gcc 3.4 32-bit asm cycles/byte
# Opteron +45% +20% 6.8 # Opteron +45% +20% 6.8
# Xeon P4 +65% +0% 9.9 # Xeon P4 +65% +0% 9.9
# Core2 +60% +10% 8.8 # Core2 +60% +10% 7.0
$output=shift; $output=shift;

View File

@ -7,9 +7,11 @@
# details see http://www.openssl.org/~appro/cryptogams/. # details see http://www.openssl.org/~appro/cryptogams/.
# ==================================================================== # ====================================================================
# SHA256 block procedure for ARMv4. # SHA256 block procedure for ARMv4. May 2007.
# Performance is ~2x better than gcc 3.4 generated code. # Performance is ~2x better than gcc 3.4 generated code and in "abso-
# lute" terms is ~2250 cycles per 64-byte block or ~35 cycles per
# byte.
$ctx="r0"; $t0="r0"; $ctx="r0"; $t0="r0";
$inp="r1"; $inp="r1";
@ -165,7 +167,7 @@ $code.=<<___;
tst lr,#1 tst lr,#1
moveq pc,lr @ be binary compatible with V4, yet moveq pc,lr @ be binary compatible with V4, yet
bx lr @ interoperable with Thumb ISA:-) bx lr @ interoperable with Thumb ISA:-)
.size sha256_block_data_order,.-sha1_block_data_order .size sha256_block_data_order,.-sha256_block_data_order
.asciz "SHA256 block transform for ARMv4, CRYPTOGAMS by <appro\@openssl.org>" .asciz "SHA256 block transform for ARMv4, CRYPTOGAMS by <appro\@openssl.org>"
___ ___

View File

@ -16,6 +16,7 @@
# -m64 -m32 | -m64 -m32 # -m64 -m32 | -m64 -m32
# --------------------------------------+----------------------- # --------------------------------------+-----------------------
# PPC970,gcc-4.0.0 +50% +38% | +40% +410%(*) # PPC970,gcc-4.0.0 +50% +38% | +40% +410%(*)
# Power6,xlc-7 +150% +90% | +100% +430%(*)
# #
# (*) 64-bit code in 32-bit application context, which actually is # (*) 64-bit code in 32-bit application context, which actually is
# on TODO list. It should be noted that for safe deployment in # on TODO list. It should be noted that for safe deployment in

View File

@ -12,7 +12,7 @@
# April 2007. # April 2007.
# #
# sha256_block_data_order is reportedly >3 times faster than gcc 3.3 # sha256_block_data_order is reportedly >3 times faster than gcc 3.3
# generated code (must to be a bug in compiler, as improvement is # generated code (must be a bug in compiler, as improvement is
# "pathologically" high, in particular in comparison to other SHA # "pathologically" high, in particular in comparison to other SHA
# modules). But the real twist is that it detects if hardware support # modules). But the real twist is that it detects if hardware support
# for SHA256 is available and in such case utilizes it. Then the # for SHA256 is available and in such case utilizes it. Then the