OpenSSL 1.0.2l release

Riccardo Mottola rmottola at
Mon Jun 5 23:32:37 CEST 2017


I actually launched a rebuild yesterday, but an incomplete reemerge, so 
I completed today.
But no issues with you fixing and completing the work, Dago!

are unstable and experimental "comparable" in terms of hardware specs? 
Perhaps the some stuff was busy or not?

I just quickly glanced over things

9s seems even a bit faster on exemperimental vs. unstable: see. eg.  8k 
md5 ahd sha1, sha512,  however aes-256 and sha256 are quite a bit 
slower, but nothing determiental.

9x seems however "broken" ? some things like des are equivalent, but md5 
and sha1 are reeeealy a lot slower:

444568.92k -> 309649.41k
412030.29k -> 277086.21k

or look at rsa verify:

5441.5 -> 2712.7

half speed! Crappy sun C ? or perhaps the difference is architecture. 
one says -march=pentium the other -xarch=generic (although I rised the 
level to pentium_pro) also it has only -O, should there be an O3?

10x however is fine... so perhaps it is not really worth to check 9x. 
Actually some stuff in 10x got faster. 11x is a bit bettere there, worse 
there and it looks like the fastest box!

Dagobert Michelsen wrote:
> As I don’t see any major flaws I’ll push the packages now.
> If you encounter anything suspicous just drop us a note.

Nothing totally bad.. perhaps we can tweak the targets a little better 
again? AS said I dumbed them down, since the optimized ones applied with 
a patch caused build failures.

> @Riccarado: Thanks a lot for all your hard work on getting this done!

no problem - we all benefit from a more up-to-date OpenSSL. Let's see if 
Jan comes up with improvements.

Looking forward getting GNUTLS working too :)


More information about the maintainers mailing list