From: Allen D. <all...@uc...> - 2004-05-11 19:43:39
|
On Tue, 11 May 2004, Scott Cain wrote: > On Tue, 2004-05-11 at 12:59, Lincoln Stein wrote: > > A cluster of 4 nodes is not really going to be adequate for decent > > cross species comparison. It would make a great test bed for a > > prototype MOD web site, though. You would designate one of the nodes > > to run the database, another to run the web site, a third to run > > BLAST, and a fourth for developing the next release. > > > > I think we can make a good argument for making the cluster into a > > hardware reference platform for a generic MOD, and skip the cross > > species comparison altogether unless someone like Jason thinks that > > we could an all-against-all BLAST across six species on a 4-node > > cluster in reasonable time. > > > > I'm willing to do the writeup if I get to have the system in my > > office. > > :-) I was thinking in my basement. I suppose it may not be heavy duty > enough to do all v all BLASTs (although, if it were BLASTP, it might > have a shot). The app could say that it could be used as test cluster > for determining needs for bigger (imply, Apple) cluster. i was thinking in our server room, we have gigabit uplink :). i agree that we can make a strong argument here for a GMOD site test bed. i do think though that lincoln would be an appropriate person to spearhead the application composition with his superb writing skillz. -allen > > Scott > > > > > Lincoln > > > > On Monday 10 May 2004 09:35 pm, Scott Cain wrote: > > > Hi, > > > > > > I'm forwarding an email from Chris Dagdigian from the bioperl > > > mailing list. I think we could put together an application for > > > GMOD to use both as a development box for GMOD applications as well > > > as a compute cluster for cross species comparisons. Would anybody > > > be willing to take the lead? The application deadline is June 13. > > > > > > Thanks, > > > Scott > > > > > > From Chris: > > > > > > Hi folks, > > > > > > Apologies for the semi off topic posting but I think it may be > > > appropriate because I know lots of people doing really clever > > > informatics stuff tend to be on the open-bio.org mailing lists. > > > > > > I want to make sure many of the small group, individuals and DIY > > > types hear about this as they can probably benefit the most from a > > > influx of hardware. > > > > > > In short, Apple just announced a program to give away 5 fully > > > provisioned mini-clusters (4 Xserve G5 nodes) in a judged process > > > specifically aimed at bioinformatics types. Awards are announced on > > > 29 June 2004. > > > > > > The URL is here: > > > http://www.apple.com/science/clusteraward/ > > > > > > Regards, > > > Chris > |