[RC5] project suggestions

david fleischer cilantro_il at yahoo.com
Tue May 22 02:07:29 EDT 2007


One interesting alternative is the Boinc client... true there is an overhead
in installing it, but that way you can concentrate on developing the cores,
whatever Dnet might decide to apply them to.

RC5 is valuable in that it provides a benchmark. Similar to what the EFF
did with their DES effort. But without a reward it feels pretty pointless.
Can you keep the key servers running anyway? 
It's a pity that RSA in their new organization have decided to terminate 
this, it provided a good excuse to run the client as "an inverted" CPU 
meter. Perhaps we can find another source of the prize to keep the effort
running.

Obviously there hasn't been a lot of progress on the key space, but this
was about to change, with the development of FPGA clients, and video
card systems. I'm wondering, what happened with all the FPGA designs 
that were purported to be in the final design stages?

Again, the important thing about RC5, in my opinion, is that it did a basic
type of research in that it was a "brute force" effort. Thus gauging the 
compute power relative to key strength rather than some more elaborate
analysis.

Perhaps a better way to use the CPU cycles, and the compute power of 
other devices, is for instance to devote the client to factoring. This is also 
a basic type of brute compute power benchmarking, but more involved in
terms of the theory.

Has the RSA factoring challenge also been terminated?

----- Original Message ----
From: Jim C. Nasby <decibel at distributed.net>
To: D.net Discussion <rc5 at lists.distributed.net>
Sent: Tuesday, May 22, 2007 12:24:23 AM
Subject: Re: [RC5] project suggestions


On Mon, May 21, 2007 at 04:38:41PM -0400, P WR wrote:
> There are many distributed computing projects that could use our help. And 
> helping them would give us a larger variety of projects to work on. And 
> even if you don't like my idea of merging with other projects, I definitely 
> think we need more short-term projects. There are many cool cryptography 
> projects that we could compute. And they wouldn't take hundreds of years to 
> complete like RC5.

I love the thought of helping other projects out, but there is a
down-side: there's a non-trivial amount of work involved in adding
projects to our network. For one thing, it means new clients, which
folks have to go and download. We'd also need to create new cores to go
in those new clients. I believe every new project requires an update to
the proxy code, and it certainly means new code for the master (although
there's been some work on allowing for multiple masters that might make
this easier). Finally there's stats, with the biggest challenge there
being "what should I measure".

Of course, all that is just a 'simple matter of code'. It's really just
a matter of getting it done. I don't want to rain on the parade, but we
just don't have a lot of time being spent on code changes. For stuff
like this to happen with any speed, it's going to take some folks (and
by folks I mean d.net participants) stepping up to the plate and getting
that work done.
-- 
Jim C. Nasby, Database Architect            decibel at distributed.net
Give your computer some brain candy! www.distributed.net Team #1828

Windows: "Where do you want to go today?"
Linux: "Where do you want to go tomorrow?"
FreeBSD: "Are you guys coming, or what?"
_______________________________________________
rc5 mailing list
rc5 at lists.distributed.net
http://lists.distributed.net/mailman/listinfo/rc5


More information about the rc5 mailing list