Subject: Re: RE: VERY slow ssh logins to uVAX
To: Aaron J. Grier <agrier@poofygoof.com>
From: Greg A. Woods <woods@weird.com>
List: port-vax
Date: 05/05/2005 20:04:30
[ On Thursday, May 5, 2005 at 15:01:40 (-0700), Aaron J. Grier wrote: ]
> Subject: Re: RE: VERY slow ssh logins to uVAX
>
> I was thinking vax is/was similar to the sparc case, in which
> recompiling libssl with different compiler flags made/makes a huge
> performance difference.
> 
>	http://mail-index.netbsd.org/port-sparc/2002/03/13/0006.html

that does clearly help, though I don't think it's anywhere near as
stunning as the difference between OpenSSH and SSH-3.x.....

> which crypto algorithms use floating point?

I don't really know for sure (esp. w.r.t. OpenSSL and OpenSSH), other
than what "egrep 'float|double' */*" shows in src/dist/openssl/crypto).
(which seems to be most of them :-)

All I know for certain is that SSH-2.x and SSH-3.x have always
exclusively used one or another of the various "bignum" or similar
integer-based math libraries.  The only use of floating point I can find
in SSH-3.x has to do with some tiny part of ASN1, as well as with timing
and statistics keeping.


> > Note that SSH-3 from ssh.com, aka pkgsrc/security/ssh2, doesn't use
> > libssl at all -- it's much faster than that!  :-)
> 
> and here I was assuming that security/ssh2 was openssh...
> 
> now where did I put my jump-to-conclusions mat?

;-)

No, openssh is, of course, pgksrc/security/openssh  :-)

You should try pkgsrc/security/ssh2 on your slow machines.

-- 
						Greg A. Woods

H:+1 416 218-0098  W:+1 416 489-5852 x122  VE3TCP  RoboHack <woods@robohack.ca>
Planix, Inc. <woods@planix.com>          Secrets of the Weird <woods@weird.com>