[4199] in cryptography@c2.net mail archive
Re: Quantum emulation
daemon@ATHENA.MIT.EDU (Daniel J. Frasnelli)
Fri Feb 12 20:26:04 1999
Date: Fri, 12 Feb 1999 20:16:00 -0500
From: "Daniel J. Frasnelli" <dfrasnel@csee.wvu.edu>
To: Lenny Foner <foner@media.mit.edu>
Cc: cryptography@c2.net
In-Reply-To: <9902122346.AA15591@out-of-band.media.mit.edu>; from Lenny Foner on Fri, Feb 12, 1999 at 06:46:53PM -0500
(Disclaimer: Off-topic, but relevant to anyone who might be interested in the
quantum sim project)
> This is quite cool.
>
> I assume this will come in handy for checking out some of the ideas in
> quantum computation and quantum crypto, long in advance of actually
> being able to build a real machine. Proofs are very handy beasts, but
> there's something to be said for tinkering around, too...
I've been active on the mailing list for a couple months or so,
mainly testing "what about this?" ideas on our big iron machines.
Quantum crypto is obviously my main motivation for involvement.
> Very nicely set up website, too.
There is a remote possibility the project will be relocated to
Europe in the next couple months. Without getting into the gory details,
a fellow wrote software to do quantum simulation on the Mac, patented it
in the U.S., and has threatened to sue all developers, users, etc. of
Open Qubit. We obviously get around him by moving the main development
tree to a location not governed by such silly laws.
> Wow. 63. And it's not exactly fast:
Multithreading has been discussed on the mailing list several
times. At this point, memory efficiency of our implementation is
a greater limit to the ability to do quantum factoring simulation
than processor speed. I am attempting to guide the main developers
towards MPICH, PVM, and various distributed shared memory packages.
> That's right, 14 seconds on my AMD-K6/2-400[1] Linux box.
Your friend may want to apply the
"yp-diagnostics_addon-pre-0.2.0rafal" patch to the source tree and
recompile, for somewhat more meaningful statistics. My system is
a mean, lean 500Mhz 21164a w/2M L3 cache and 128M ram. Because the
current implementation is bottlenecked at physical memory size,
the largest number I can factor is below 130. I am, however,
able to factor 111 in 3 minutes, 19 seconds (keep in mind that
Shor's algorithm assumes exponential complexity)). Not too shabby,
but I'd love to stick more memory in and see what she can really do.
Best regards and hope to see you on the Open Qubit list,
Daniel
--
