I started up a 'ps' nbody search. Let me know if you're having any problems with these workunits.

---

no problem with units..but time to completion is way off. 52 hours = 20 minutes on nvidia gts 220 amd 64bit 9750 processor ubuntu

In reguard to the above....never be in a hurry to post....that was mis-info.

Got one of the ps_nbody_test3.... tasks

Got an computation error after one second.

It's an I7 with 64bit Win 7 Ultimate. GTX460, Boinc 6.12.26


Status of the task is running, despite it quit.
ps_nbody_test3_70386_0
Task 42269779

Just another edit:

Got a second one on another machine. It was successful. Runtime 9:06 on 5 CPU's
Also I7 with 64bit Win7 Ultimate. GTS240 Boinc 6.12.26
ps_nbody_test3_54182_2
Task 42051075

---

Got two more error tasks on two machines

This is the outputfile. Its the same output at every error task.

<core_client_version>6.12.26</core_client_version>
<![CDATA[
<message>
Unzul�ssige Funktion. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
Error evaluating script: [string "..."]:33: bad argument #1 to 'plummerTimestepIntegral' (Invalid small radius)
Error loading Lua script 'nbody_parameters.lua': [string "..."]:33: bad argument #1 to 'plummerTimestepIntegral' (Invalid small radius)
Failed to read input parameters file
Failed to setup run
<search_application>milkywayathome nbody 0.62 Windows x86_64 double OpenMP Crlibm</search_application>
12:37:08 (9548): called boinc_finish

</stderr_txt>
]]>

Got ps work unit and it showed over 120 hours with a short time for completion, about 10 days. That makes it hog CPU and it's currently running high priority, using both CPUs instead of sharing one with other projects. I don't mind work units that take a long time, but don't appreciate CPU hogging. Need to set a longer deadline for such long WUs.

The nbody's for me are estimated near 600 hours and finish in around 3. However the ps seem to be a really close estimate for the runtime.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Estimate on ps seems to be close, judging by the amount of work finished and left to go. The WU has been using both processors every time I've looked over the last 24 hours. I've had nbody WUs in the recent past that had long completion estimates and completed quickly, not so with this ps WU. So far, elapsed time is just over 33 hours, with just under 95 hours to go. Completion date is 7/12 at 3:33 in the afternoon.

Update on ps_nbody_test3_72292_0

Still hogging CPU. Shows 42:39:24 elapsed and 89:07:06 to go. It is apparently allowing other WUs that will be pushing their deadline to complete, but has both CPUs the rest of the time. Overall time for completion (adding elapsed and to go) is increasing. Started at just over 120 hours and is now over 131.

I have read in other threat but can't find it anymore, was mentioned that the feature of the n-Body was that it uses all the available cpu-cores.
I did a few on my i7 and it was using 6 cores. Calculated time in BOINC was then 3 hours but it took only 12 minutes.

---

Greetings from,
TJ

Would be fine to use all CPUs, *IF* it weren't using them all the time. It is continuing to run as high priority, preventing other WUs from running. It has been running as high priority for at least 2 days now.

I realize that, in the past, nbody WUs reported high numbers for completion time and completed much sooner - this puppy is a whole 'nuther animal. The time that it is going to take to complete will be longer than that which was originally reported, so far around 10 hours longer. That was calculated based on time used and estimated time to completion. Original time was a little over 120 hours and current calculation shows a little over 132 hours - just over 48 hours already run and estimating a little over 84 hours to go.

My problem with this WU is that the deadline is set way too short - was about 10 days from receipt of the WU, IIRC. I run climate prediction WUs so I'm not adverse to running long ones, but the deadline needs to be set to something reasonable, which won't force constant high priority.

I intend to let this one complete, but if short deadlines continue, I'll have to drop the project. I won't let one project hog the CPU (high priority all the time) at the expense of the rest - it's just not fair to the other projects I support.

Up until the ps searches came around, my "statistics" graph was rising on a pretty consistent line, averaging about 800/day. However, since working the PS units, the graph is looking like the EKG of someone about to leave. The graph is now showing an increase of just 2000 in the last 6 days. There also seems be be zero consistency when it comes to the number of processors used, the time it takes to complete a unit, or what order the units are worked.

I finished one of those long running tast while the server was down and it gives a really low amount of credit compared to other workunits. Just look at this
62821399 42326561 285968 1 Jul 2011 | 19:22:53 UTC 14 Jul 2011 | 17:18:24 UTC Fertig und Bestätigt 4,513.32 9,002.14 798.00 MilkyWay@Home N-Body Simulation v0.62 (mt)
62764344 41506737 285968 1 Jul 2011 | 17:26:50 UTC 14 Jul 2011 | 17:18:24 UTC Fertig und Bestätigt 543,932.54 1,058,716.00 2,500.00 MilkyWay@Home N-Body Simulation v0.62 (mt)
and I havent crunched anything else the previous ten days.