[hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- From: Jonathan Morton <chromi@xxxxxxxxxxxxxxxxxxxxx>
- To: hashcash@xxxxxxxxxxxxx
- Date: Thu, 5 Aug 2004 17:04:16 +0100
CPU Info I was getting 1.25 MegaSHA1/sec using only ANSI-C:
processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 6
model name : AMD Athlon(tm) XP 1700+
stepping : 2
cpu MHz : 1460.471
cache size : 256 KB
I'm getting 2 MegaSHA1/sec using three different ANSI C cores on a
1600MHz Athlon-XP, and 3 MegaSHA1/sec using the MMX version. This is
when optimised for a Pentium-MMX, not an Athlon. A 3.06GHz P4 has been
reported as breaking 4 MegaSHA1/sec. We don't yet have figures for a
K8 class chip.
On a 667MHz G4, I'm getting up to 1.2 MegaSHA1/sec using an ANSI C core
(if the compiler options are very carefully controlled), and 3
MegaSHA1/sec using Altivec. Extrapolating, a 2.0 GHz G5 might
conceivably reach 9 or more, assuming everything remains
single-threaded.
Performance isn't really a concern any more - what we need is stability
and out-of-the-box compatibility with old chips. I've thought up a
good way to handle the latter (simply catch the SIGILL that CPUID might
trigger), so if the Debian folks would like to just make an i586
package for now, I'll come forward with an i386-compatible version in a
while.
Right now I'm also working on getting the minter to actually work with
my code integrated. The first step is to make it compile within my
preferred devel environment. Until that happens, the Debian folks
should prepare a package *without* my code integrated. It'll be
slower, but slow is better than non-functional. If that means the fast
code gets bumped to the unstable tree due to missing the deadline, so
be it.
--------------------------------------------------------------
from: Jonathan "Chromatix" Morton
mail: chromi@xxxxxxxxxxxxxxxxxxxxx
website: http://www.chromatix.uklinux.net/
tagline: The key to knowledge is not to rely on people to teach you it.
Other related posts:
- » [hashcash] cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)
- » [hashcash] Re: cpu detection, disabling mmx code on non mmx cpu (Re: Re: status of hashcash version 1?)