I am releasing a new version of N-Body with some MINOR changes. There should be nothing different from 1.38 in terms of the cruncher's side. However, if you notice any problems, please post them here.

Jake

---

I have only had one ver 1.40 task that is not mt but that's fine by me. I've never run n-body before so these are a bit of a novelty. Since I only run one CPU project at a time, there are no conflicts.

I seem to get a lot of N-Body Simulation 1.40 (mt) wu that have 1-2 hour time estimates, then about an hour in, the time remaining starts counting up. Most of these take anywhere from 6 hours to 15 hours to actually complete. They report using 2 or 4 cpu (which is correct for each). I realize that it is an estimate, however I don't believe it is calculating the estimate anywhere near reality. I see this on two machines both running Win 7 (64bit), Boinc 7.2.33 one with Intel Core 2 Duo 2.66 and one with Core I5 3.1 (oc to 3.7) both crunching multiple projects but only one project at a time.

Jake, when are we going to get the non-mt WU?

Long mt-WUs tax the poor little Boinc brain....

I am using Boinc 7.2.7 (x64) on Linux Mint version 14 (Nadia). I am getting a copy of n-body 1.4 marked "Ready to start" for each core on the computer, except that they never start. I am experiencing the same thing on each of 3 separate machines, 2 of which run AMD multi-core processors, and one runs an Intel dual. I would send you a copy of the properties panel if Boinc would allow me to copy text from it. Does this problem ring any bells? Alternatively, is there any debugging info I can easily access and send?

Thanks

Under "Tools - Computing Preferences...", near the bottom, what is your "On Multiprocessor Systems Use At Most xxx processors" set to? Also, if you click properties on your task, how many cpu cores are those tasks set to use?

I seem to get a lot of N-Body Simulation 1.40 (mt) wu that have 1-2 hour time estimates, then about an hour in, the time remaining starts counting up. Most of these take anywhere from 6 hours to 15 hours to actually complete. They report using 2 or 4 cpu (which is correct for each). I realize that it is an estimate, however I don't believe it is calculating the estimate anywhere near reality. I see this on two machines both running Win 7 (64bit), Boinc 7.2.33 one with Intel Core 2 Duo 2.66 and one with Core I5 3.1 (oc to 3.7) both crunching multiple projects but only one project at a time.

I have similar issues. A recent task came in at over 6mil gigaflops. Somehow now it changed to 300k or so gigaflops under properties. Been going for over 5 hrs and finally half way done. Clock counting up on time left estimating about and hour now but obviously it will take it another 5 if it's only 50% now.

Under "Tools - Computing Preferences...", near the bottom, what is your "On Multiprocessor Systems Use At Most xxx processors" set to? Also, if you click properties on your task, how many cpu cores are those tasks set to use?

The XXX value is 0.00, which I assume means 100% by default. On each system, the number of cores set per task is that of the processor on the system, namely 2 for the Intel, and 4 & 8 for the AMDs.

I have gotten the N-Body Simulation 1.40 (mt) tasks running successfully on Linux Mint 16. The issue is that I had not previously selected N-Body Simulations on my account with MilkyWay. Why are these running when I had NOT selected them to run????????

We need to have a way to SHARE computer resources. Turning mt on and off selectively to allow other BOINC projects to run should be something that we are allowed to do.

Right now, MW is downloading N-Body Simulations in conflict with my selections on the web site.

Do you have this box checked?

If no work for selected applications is available, accept work from other applications?