[rc5] distributed.net: a Generalised Compute Resource?

Henry W Miller mill0440 at gold.tc.umn.edu
Fri Jun 20 14:41:35 EDT 1997



On Fri, 20 Jun 1997, Matt J. Gumbley wrote:

> The idea of a "generalised compute resource" is very interesting: that the
> machines we currently use give up some cycles to *whatever job we throw at
> them*.
> 
> Here's a (rather rambling) Friday-afternoon thought for you:
> 
> Each machine runs a shell. This provides the following:
> 1) The user interface, a la Bovine Win32 client, perhaps.
> 2) The network I/O connection to the proxies.

<snip>

There needs to be a little more.  Many people (and I don't blame them) do
not want the general client downloading unknown binaries.  The shell
should provid for people not wanting to download binaries.  

Also we need to make sure the platform is unique enoguh that differences
can be handeled.  For instance NetBSD/sparc can run solaris binaries, but
most people don't desire to do so. (I think there is a proformance hit)
a.out binaries won't run on all Linux elf systems (mine for one) and vis
versa.  There are PPC and PowerII RS/6000 machines.  OS/2 can run windows
programs, but some people (again like me) don't hvae that support
installed, even if they do there is a proformance hit.

Also Sparc is not as good for RC5 as other chips.  It is better use fo
reasources to run the sparcs on the primes problem.  But some people may
want to try their hand at rc5 anyway, so they need to have that option to
force rc5.

You did have some good ideas though.  Specifics should be discussed
elsewhere, in part because obscurity is good secerity, though not perfect.

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list