- Apr 7, 2003
- 4,278
- 0
- 0
Originally posted by: MrSpadge
Interesting system, interesting thread!
One idea which I have in mind after reading through it: from what you say it sounds like you have to wait for the results of the computations until you can continue working? And therefore you need maximum performance in one box, because the software is multi threaded but the tasks can not be split among several loosely coupled machines? (i.e. a cluster with GBit ethernet)
If I'm wrong in any of the points above you could do all kinds of funny things like ordering 3 boxes with 2 2GHz quads each (for the pure calculations) or you could use 2 machines in parallel for the editing, i.e. your old one and a new one or 2 somewhat smaller new ones.
Regards, MrS
PS @Idontcare & derwen: nice to see an argument being settled like it should be done among grown-up people. A refreshing change![]()
I'm pretty sure the software isn't set up to run on a "cluster" per-se, nor would I know how to set that up with this software. Any of the really long runs are done on the mainframe so I'm never waiting more than an hour for my own calculations to run and I usually do as much optimization as I can in a night then toss my run in the que just before I leave in the morning and when I get back that night it's almost always done. Can't hurt to ask though right? I could always go the way of the USAF and just order a bunch of PS3's.
