 |
| [OCAU] badger 2004-05-30 21:23:12 | I just watched a solenoids to 15cm sim on the commandline (I usually run it in background mode). (running one last sim of 4.41 before change to 4.41a) Previous versions had a cutoff, the simulation would end at 300ns. I had noticed a little while ago that this had increased to 500 or 700ns, but today the run i watched kept going until the last perticle fell off it's perch at 12199.45ns! (this particle is no more, it has ceased to be, bereft of life he rests in peace). btw the score was 10.278729 (299.6 Mpts) Is this now the norm? Do all sims run until all particles have gone? If so multi-threading is a waste as 1/4 of the sim was less than 100 particles... |
| Herb[Romulus2] 2004-05-30 23:09:22 | That might explain, why 2 independent instances are more effective on a HT-Box |
| [TA]z 2004-05-30 23:55:25 | I'm confused as to which client you're running - I could see this happening on a SolenoidsTo15cm in v4.41, but not v4.41a... It was a bug in v4.41 and has been fixed in the latest client release. |
| Herb[Romulus2] 2004-05-31 05:06:43 | You might be right, I've seen that several times on 4.41 and now on 4.41a the last interval saved I see 500ns with a few hundred particles. |
| [OCAU] badger 2004-05-31 23:16:59 | quote: you are correct, I measured a 10% increase on a 3.0GHz HT P4 running 2 instances using v4.34 I noticed that this simulation seemed to run in spurts, about 3/4 second apart,although that could just be the update of the commandline interface. [TA]z - I was running 4.41, waiting for the sim to finish before switching to 4.41a |
