Thank you! Does this mean that it is automatically downloaded and that I will be not have to take action in the future? I appreciate the help!

Yep, as long as you run the project in a "stock" configuration (ie no app_info file), then the project server and BOINC will take care of making sure you have what you need.

Be advised though, these current nBody runs are Beta testing, so you should keep an eye on the host (and the wingmen running the nBody WU's with you) generally to make sure it is still running the app properly. There are still some problems which could result in your host not being able to run the app at all (doesn't seem to be the case for yours), or wasting a lot of it's time on tasks that may or may not complete and/or validate.

HTH

I have just compiled the static libraries for linux 64 bit.

I released them as version 1.09. Linux users may want to update.

I am adding them as part of the 1.08 release stream and any issues can be posted to this thread.


Jeff Thompson

Thanks for the static libraries... my computer just picked one up and I'm half way done it after 1/2 hour, but it's using the CPUs, not the GPU, even though it reports as a "1.09 (opencl_nvidia)" task.

Yes currently all nbodies are cpu only.

We are hoping to change this in future releases.

Though depending on the release schedule it may not be that the next releases have gpu support.

So we are also looking at if we can deprecate the current gpu labeled binaries without affecting the current runs and users.


Thanks,
Jeff Thompson

FWIW, I've been noticing that several n-bodies will be running simultaneously, and equally, at greater than 100% cpu capacity but only one will indicate "High Priority" in the Task list.

Also, I've noticed a significant drop in my daily credit average (roughly 23,000 to roughly 16,000) since the implementation of the 1.08's. Is this due to a poorer cpu usage scheme (read: efficiency) or has the credit allocation method been altered?

:)

Thank you Jeff. I will download some workunits tonight and let you know if there's any problem.

I'm running N-Body Simulation 1.08, after running for 2.5 hours the remaining time is 642.5 hours. Is there an issue here?

---

Precise estimation is nearly impossible here.

This is probably due to nature and complexiTY of the NBody algorhythm...

It is heavily using 23 MegaBytes of memory and so the speed may vary from 1 unit (hour) to 400 times that unit (1 hour)... - depending on this:
If all the needed data fits into the cache of that CPU in time and in advance (thanx to Out-Of-Order-Execution feature in modern Intel CPUs...) , for example into 512 MB per core + 6 MB L3 cache in core i7-720QM , it will run 400 times faster BECAUSE:

RAM is generally about 400 times slower than cache and CPU itself !


May be the estimation does NOT count on this - that it will always fit into cache... on all CPUs - and that is of course true. But it WILL fit on most of them.

So, it really depends whether it fits into cache or NOT. if it HAS TO frequently wait 400 cycles for a RAM-access read of data, that estimation would be CORRECT.

---

Melwen - Child of the Fangorn Forest
Rig "BRISINGR" [ASUS G73-JH, i7 720QM 1.73, 4x2GB DDR3 1333 CL7, ATi HD5870M 1GB GDDR5],bought on 2011-02-24

I have a bunch of extremely-fast NBody 1.08 WUs here lately.
They complete in about 6 minutes !

Are they OK ??? Is it normal ?

For example this unit:
http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=355519412

---

Melwen - Child of the Fangorn Forest
Rig "BRISINGR" [ASUS G73-JH, i7 720QM 1.73, 4x2GB DDR3 1333 CL7, ATi HD5870M 1GB GDDR5],bought on 2011-02-24

And this WU: http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=356446450

---

Melwen - Child of the Fangorn Forest
Rig "BRISINGR" [ASUS G73-JH, i7 720QM 1.73, 4x2GB DDR3 1333 CL7, ATi HD5870M 1GB GDDR5],bought on 2011-02-24

I have got too many computational errors lately on CPU AMD A8-3870K running Win8 x64.

For example WU id 356842510: http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=356842510

***** --- See? This is ridiculous ! Seems like the error ocurred in the time when all computations should be successfully completed (given the time in seconds it was running)! - just in the moment when it (BOINC?) was trying to *SEND* the resulting file (maybe too *early* when the computing-app was still in the process of shutting down and releasing its *LOCK* on the file it has just written on disk - using some WINDOWS-system call of course... ).


I think I will have to switch this PC to compute another project (for example SETi) which will be more effective (no errors).

WHAT IS THIS ERROR ANYWAY ???


Error 4 (0x4) - Unknown error number

CODE of the output:

<core_client_version>7.0.64</core_client_version>
<![CDATA[
<message>
The system cannot open the file.
 (0x4) - exit code 4 (0x4)
</message>
<stderr_txt>
<search_application> milkyway_nbody 1.08 Windows x86_64 double  OpenMP, Crlibm </search_application>
Using OpenMP 1 max threads on a system with 4 processors
Using OpenMP 1 max threads on a system with 4 processors
tree-incest detected (fatal) at step 6494 / 7014 (92.586256%)
Error running system: NBODY_TREE_INCEST_FATAL (16)
11:51:17 (4656): called boinc_finish

</stderr_txt>
]]>

"The system cannot open the file." ???


Impossible! - No other process nor Windows 8 would ever lock a file or partition table... Are these applications: BOINC (when downloading a WU) and NBody - both taking into account the new file-system-related user/process rights in WINDOWS 8 ??? Does anybody have the same problem on Windows 8 x64 Prof. (or Preview release) ?

Otherwise this does not make sense to me. The disk is ok.
... OK, I will scan it for logical errors every day, just for sure.
I will also try to reset the project... and I will report back *IF* the problem persists.

Reason of edit: to add link to exemplary WU.
Reason of edit: *added* "SEEMS LIKE" ...: "Seems like the error ocurred" - because there is also the *OTHER* error that ocurred in about 92.5 percent the NBODY_TREE_INCEST_FATAL.

---

Melwen - Child of the Fangorn Forest
Rig "BRISINGR" [ASUS G73-JH, i7 720QM 1.73, 4x2GB DDR3 1333 CL7, ATi HD5870M 1GB GDDR5],bought on 2011-02-24

FWIW, I've been noticing that several n-bodies will be running simultaneously

Three simultaneous tasks crunching here...

Windows 7 64-bit, BOINC manager 7.0.64 (x64), MilkyWay@Home N-Body Simulation 1.08
Intel Core i7 (8 core) CPU

N-Body Simulation 1.08 ps_nbody_100k_chisq_alt_40913_1366886102_450601_2

has a run time of 280 hours. ..._441576_3, ...458261_0, ... 457251_1 also

had really long run times. it that normal???

---

now that it's been running for two hours the remailing time has changed to 17 hours

Most files of this nature will show bizarre estimates when run for the first few times - the calculations are so complex, they have little to go on to make an estimate the first few times around.

As a (very) crude generalisation - once they get to around 30% complete the estimates are usually pretty good, after 60% done its pretty well nailed. Cant get specific, it varies hugely depending on whats being crunched, but that's a reasonable guideline.

any chance that gpu side could have single persion? LOL