On Tue, Mar 05, 2002 at 11:47:16PM +0100, Martin Husemann wrote:
> > I see a long delay when running ssh to a 486/50 I have lying 
> > around, but this sounds worse.
> 
> I see the same symptoms. I can ssh (protocol v2 used) to a 486/50, and
> it will take quite some time. I can ssh (protocol v2 too) to a LX or
> an SS2 and it will take at least double the time (if not more).
> 
> No IPv6 involved, no DNS issues.
> 
> I always thought this would be due to some assembly stubs used in the
> i386 version of ssh/libssl.

the problem has shown up in other ports (68k and vax).  based on my
limited testing, I posit it's an optimization problem.

there's a speed benchmark program (speed.c) in
src/crypto/dist/openssl/apps if anybody else wants to take a closer
look, or verify my results.

this is my sparc 2:

OpenSSL 0.9.6b 9 Jul 2001
built on: Fri Jan 18 09:30:42 MET 2002
options:bn(32,32) md2(int) rc4(ptr,int) des(idx,cisc,16,long)
blowfish(ptr) 
compiler: /usr/tools/bin/sparc--netbsdelf-gcc -O2 -pipe   -Werror
                  sign    verify    sign/s verify/s
dsa  512 bits   0.5157s   0.6235s      1.9      1.6
dsa 1024 bits   1.8048s   2.2893s      0.6      0.4

and this is my decstation 5000/240:

OpenSSL 0.9.6 24 Sep 2000
built on: Wed Jun  6 09:01:40 PDT 2001
options:bn(64,32) md2(char) rc4(idx,int) des(idx,cisc,16,long) idea(int)
blowfish(idx) 
compiler: gcc -fPIC -DPIC -DTERMIOS -O3 -Wall -DL_ENDIAN
                  sign    verify    sign/s verify/s
dsa  512 bits   0.0826s   0.1048s     12.1      9.5
dsa 1024 bits   0.2634s   0.3280s      3.8      3.0

both are 40MHz processors, yet MIPS is almost 10 times faster than SPARC
here.

-- 
  Aaron J. Grier | "Not your ordinary poofy goof." | agrier@poofygoof.com
      "[...] I generally haven't found IDM guys to be very good
       live acts, most of them just sit down at their laptop and
       tweak reaktor."  -- Brandon Daniel