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

sha/asm/sha*-mb-x86_64.pl: commentary update.

Browse Source
This commit is contained in:
Andy Polyakov 2013-10-08 23:36:55 +02:00
parent c45a48c186
commit 61ba602af5
2 changed files with 18 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
crypto/sha/asm/sha1-mb-x86_64.pl
View File

@ -12,20 +12,22 @@
# naturally limited to 4 on pre-AVX2 processors and to 8 on
# AVX2-capable processors such as Haswell.
#
# this +aesni(*) sha1 aesni-sha1 gain
# this +aesni(i) sha1 aesni-sha1 gain(iv)
# -------------------------------------------------------------------
# Westmere(**) 10.4/n +1.28=3.88(n=4) 5.44 6.58 +70%
# Atom(**) 18.9/n +3.93=8.66(n=4) 10.0 14.0 +62%
# Westmere(ii) 10.4/n +1.28=3.88(n=4) 5.44 6.58 +70%
# Atom(ii) 18.9/n +3.93=8.66(n=4) 10.0 14.0 +62%
# Sandy Bridge (8.16 +5.15=13.3)/n 4.99 5.98 +80%
# Ivy Bridge (8.03 +5.14=13.2)/n 4.60 5.54 +68%
# Haswell(***) (8.96 +5.00=14.0)/n 3.57 4.55 +160%
# Haswell(iii) (8.96 +5.00=14.0)/n 3.57 4.55 +160%
# Bulldozer (9.75 +5.76=15.5)/n 5.95 6.37 +64%
#
# (*) multi-block CBC encrypt with 128-bit key;
# (**) (HASH+AES)/n does not apply to Westmere for n>3 and Atom,
# (i) multi-block CBC encrypt with 128-bit key;
# (ii) (HASH+AES)/n does not apply to Westmere for n>3 and Atom,
# because of lower AES-NI instruction throughput;
# (***) "this" is for n=8, when we gather twice as much data, result
# (iii) "this" is for n=8, when we gather twice as much data, result
# for n=4 is 7.98+4.44=12.4;
# (iv) improvement coefficients in real-life application are somewhat
# lower and range from 30% to 100% (on Haswell);
$flavour = shift;
$output = shift;

16
crypto/sha/asm/sha256-mb-x86_64.pl
View File

@ -12,21 +12,23 @@
# naturally limited to 4 on pre-AVX2 processors and to 8 on
# AVX2-capable processors such as Haswell.
#
# this +aesni(*) sha256 aesni-sha256 gain
# this +aesni(i) sha256 aesni-sha256 gain(iv)
# -------------------------------------------------------------------
# Westmere(**) 23.3/n +1.28=7.11(n=4) 12.3 +3.75=16.1 +126%
# Atom(**) 39.1/n +3.93=13.7(n=4) 20.8 +5.69=26.5 +93%
# Westmere(ii) 23.3/n +1.28=7.11(n=4) 12.3 +3.75=16.1 +126%
# Atom(ii) 39.1/n +3.93=13.7(n=4) 20.8 +5.69=26.5 +93%
# Sandy Bridge (20.5 +5.15=25.7)/n 11.6 13.0 +103%
# Ivy Bridge (20.4 +5.14=25.5)/n 10.3 11.6 +82%
# Haswell(***) (21.0 +5.00=26.0)/n 7.80 8.79 +170%
# Haswell(iii) (21.0 +5.00=26.0)/n 7.80 8.79 +170%
# Bulldozer (21.6 +5.76=27.4)/n 13.6 13.7 +100%
#
# (*) multi-block CBC encrypt with 128-bit key;
# (**) (HASH+AES)/n does not apply to Westmere for n>3 and Atom,
# (i) multi-block CBC encrypt with 128-bit key;
# (ii) (HASH+AES)/n does not apply to Westmere for n>3 and Atom,
# because of lower AES-NI instruction throughput, nor is there
# AES-NI-SHA256 stitch for these processors;
# (***) "this" is for n=8, when we gather twice as much data, result
# (iii) "this" is for n=8, when we gather twice as much data, result
# for n=4 is 20.3+4.44=24.7;
# (iv) improvement coefficients in real-life application are somewhat
# lower and range from 75% to 130% (on Haswell);
$flavour = shift;
$output = shift;