I've had the same experience with Virtual Prairie. Two instances is the most that my quads will run at a time efficiently. The other 2 cores will run a different project such as Spinhenge or PrimeGrid very well while those 2 instances of Virtual Prairie crunch.
So off the top of my head we have Lattice, NFS, CPDN, CPDNbeta, and Virtual Prairie that all need to be run with fewer than 4 instances on a quad core if any kind of efficiency is to be maintained. Amount of RAM is not the issue, seems to be more an issue of the size of the working set and swapping in and out of main memory. That's a quick 5 projects, I'm sure there are more. Regards/Ed On Fri, Feb 12, 2010 at 12:30 PM, Uwe Herzke <[email protected]> wrote: > I don't have "operator knowledge", but reasonable insight what's running > good on my machine and what not. > > I know, that a single Climate is more effective than 2 in parallel, I > get this set-up by setting CPDN to NNW after every DL and manually look > after the WU once it's finished. As they take quite long it's not that > big effort, but an automatism would be fine. > > I know that I must not crunch more than 2 WUs of VirtualPrairie in > parallel, they get veeeryyy slooowww in this case. This is a task that > can only be done properly with very much fiddling with stopping bunches > of WUs, setting NNW, constantly babysitting the tasks tap. Otherwise the > WU will take double as much time as needed, a big waste of computation > power. > > I'd very much like an option to set each project to a max. number of > parallel running WUs. It will probably be 4 for most projects on my > quad, but not for CPDN, CPDNbeta, VirtualPrairie and probably more in > the future. > > Greets, Saenger > > > Original message: > > If "operator knowledge" is required, then BOINC is not "set and forget." > > > > ... and if the majority of BOINC users are "set and forget" that has > > to drive the feature set. > > > >> On 2/12/2010 7:36 AM, Raistmer wrote: > >> > >> Actually I don't think it's really complex. All needed is operator > >> knowledge > >> what app performs better paired with what another app. > >> > >> This knowledge can be aquired from simple experiments > _______________________________________________ > > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
