summaryrefslogtreecommitdiff
path: root/src/lib/libcrypto
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--src/lib/libcrypto/des/times/usparc.cc2
-rw-r--r--src/lib/libcrypto/ripemd/README2
2 files changed, 2 insertions, 2 deletions
diff --git a/src/lib/libcrypto/des/times/usparc.cc b/src/lib/libcrypto/des/times/usparc.cc
index f6ec8e8831..0864285ef6 100644
--- a/src/lib/libcrypto/des/times/usparc.cc
+++ b/src/lib/libcrypto/des/times/usparc.cc
@@ -2,7 +2,7 @@ solaris 2.5.1 usparc 167mhz?? - SC4.0 cc -fast -Xa -xO5
2 2
3For the ultra sparc, SunC 4.0 cc -fast -Xa -xO5, running 'des_opts' 3For the ultra sparc, SunC 4.0 cc -fast -Xa -xO5, running 'des_opts'
4gives a speed of 475,000 des/s while 'speed' gives 417,000 des/s. 4gives a speed of 475,000 des/s while 'speed' gives 417,000 des/s.
5I belive the difference is tied up in optimisation that the compiler 5I believe the difference is tied up in optimisation that the compiler
6is able to perform when the code is 'inlined'. For 'speed', the DES 6is able to perform when the code is 'inlined'. For 'speed', the DES
7routines are being linked from a library. I'll record the higher 7routines are being linked from a library. I'll record the higher
8speed since if performance is everything, you can always inline 8speed since if performance is everything, you can always inline
diff --git a/src/lib/libcrypto/ripemd/README b/src/lib/libcrypto/ripemd/README
index 7097707264..f1ffc8b134 100644
--- a/src/lib/libcrypto/ripemd/README
+++ b/src/lib/libcrypto/ripemd/README
@@ -4,7 +4,7 @@ http://www.esat.kuleuven.ac.be/~bosselae/ripemd160.html
4This is my implementation of RIPEMD-160. The pentium assember is a little 4This is my implementation of RIPEMD-160. The pentium assember is a little
5off the pace since I only get 1050 cycles, while the best is 1013. 5off the pace since I only get 1050 cycles, while the best is 1013.
6I have a few ideas for how to get another 20 or so cycles, but at 6I have a few ideas for how to get another 20 or so cycles, but at
7this point I will not bother right now. I belive the trick will be 7this point I will not bother right now. I believe the trick will be
8to remove my 'copy X array onto stack' until inside the RIP1() finctions the 8to remove my 'copy X array onto stack' until inside the RIP1() finctions the
9first time round. To do this I need another register and will only have one 9first time round. To do this I need another register and will only have one
10temporary one. A bit tricky.... I can also cleanup the saving of the 5 words 10temporary one. A bit tricky.... I can also cleanup the saving of the 5 words