You might consider leaving hyperthreading enabled and just run 2 instances on each physical processor. On my dual 2.8 GB P4 Xeon (running Win2K), I found that the virtual CPU's were able to run a lot of the ordinary OS tasks, leaving the physical ones with even more time to do FPU crunching.
And yeah, boy oh boy they could burn through an exponent quickly! I wish I had a room full of 'em. :) > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of > John R Pierce > Sent: Friday, March 14, 2003 5:08 PM > To: Mersenne discussion list > Subject: Mersenne: p4 xeons... > > > oh boy, maybe I can climb back up the ranks a bit having long > since slipped > off the first 100... I just brought online a pair of identical Intel > servers, each a 2GB dual p4/xeon 2.8Ghz linux 2.4.18 > system... running latest non-beta mprime from the downloads page ... > > I have disabled hyperthreading on these two servers, so they > each look like 2 CPUs rather than 4. Looks like each CPU > thinks it can complete a 18,600,000 sized assignment in 10 days. > > one minor mprime question... the readme implied that if I > specified `mprime -b2` it would spawn two copies, but it only > seemed to spawn one... I had to run `mprime -b` and `mprime > -b2` to get both CPUs 100% busy (after manually setting up > primenet by using `mprime -m` and `mprime -m -a1` ... > > another minor question... Is there any way to force CPU > affinity, or does mprime do that automatically? _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
