Welcome to MilkyWay@home

Posts by Brian Priebe

1) Message boards : Number crunching : Errors (Message 66612)
Posted 16 Sep 2017 by Brian Priebe
Post:
I also just got 40+ failed WU's in a row. All of them failed on the same error:

<number_WUs> 5 </number_WUs>
<number_params_per_WU> 21 </number_params_per_WU>
Number of parameters doesn't make sense
2) Message boards : Number crunching : Cannot Upload Results (Message 66233)
Posted 20 Mar 2017 by Brian Priebe
Post:
20-Mar-2017 05:58:29 | Milkyway@Home | Sending scheduler request: Requested by user.
20-Mar-2017 05:58:29 | Milkyway@Home | Reporting 25 completed tasks
20-Mar-2017 05:58:29 | Milkyway@Home | Requesting new tasks for AMD/ATI GPU
20-Mar-2017 05:58:30 | Milkyway@Home | Scheduler request completed: got 0 new tasks
20-Mar-2017 05:58:30 | Milkyway@Home | Server can't open database
3) Message boards : Number crunching : Erroneous WU Download for Non-DP GPU (Message 66029)
Posted 19 Dec 2016 by Brian Priebe
Post:
This needs to be excluded on only one of a handful of hosts.

MilkyWay GPU detection should surely handle this situation automatically.
4) Message boards : Number crunching : Erroneous WU Download for Non-DP GPU (Message 66026)
Posted 18 Dec 2016 by Brian Priebe
Post:
After receiving over 400 GPU tasks, I have suspended this project on that machine.
5) Message boards : Number crunching : Erroneous WU Download for Non-DP GPU (Message 66015)
Posted 14 Dec 2016 by Brian Priebe
Post:
Just installed BOINC 7.6.33 on a domain controller with an AMD 6850 GPU. When I joined it to MilkyWay, BOINC downloaded a plethora of GPU work units for some reason. Naturally they all failed for lack of double precision. Is this normal? What needs to be done to prevent it from attempting more GPU work units?
6) Message boards : Number crunching : AMD Bonaire Support? (Message 63873)
Posted 13 Aug 2015 by Brian Priebe
Post:
Are there any plans to issue GPU work to AMD Bonaire cards any time soon? I notice this was discussed a year ago...
7) Message boards : Number crunching : Finish File Present Too Long? (Message 63360)
Posted 12 Apr 2015 by Brian Priebe
Post:
Out of the blue, I had two GPU WU's fail. One example is http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1074095584. Both WU's have the same symptoms as listed below.

The error code is listed as: 194 (0xc2) EXIT_ABORTED_BY_CLIENT

In the stderr output is the odd notation:

<message>
finish file present too long
</message>


Any ideas what causes this?
8) Message boards : Number crunching : Benchmark results - times wanted for any hardware, CPU or GPU, old or new! (Message 61091)
Posted 12 Feb 2014 by Brian Priebe
Post:
Actually my bad. Didn't see the 213 credit notice for some reason. All of them were only 159.86 credits. There is nothing listed recently @213 credits.
9) Message boards : Number crunching : Benchmark results - times wanted for any hardware, CPU or GPU, old or new! (Message 61057)
Posted 11 Feb 2014 by Brian Priebe
Post:
With reference to computer ID 313473 with a dead stock AMD HD7970 GPU @925Mhz (1375Mhz memory clock):

Most recent 260 WU's are minimum 61.56sec, maximum 68.12sec, average 63.15sec. Only one WU executed at a time.
10) Message boards : Number crunching : GPU apps delivered to single precision GPU (Message 58591)
Posted 9 Jun 2013 by Brian Priebe
Post:
Lately I am seeing with this GPU app downgrade to 0.82 that the server is all of a sudden delivering GPU WU's to systems without double precision capability. What's up with that?
11) Message boards : News : New Separation Run Started - ps_p_80_3s_dr8_4 (Message 57907)
Posted 12 Apr 2013 by Brian Priebe
Post:
Boinc 7.0.28 is the problem. If you upgrade to a newer version, your issue will go away.

Where do we get the newer version? And which newer version? The last production version posted at BOINC.BERKELEY.EDU is 7.0.28. 7.0.62 is marked for testing only.
12) Message boards : Number crunching : Invalid XML in project preferences? (Message 55381)
Posted 18 Aug 2012 by Brian Priebe
Post:
All of my work units are showing this in the output now:

Unrecognized XML in project preferences: nvidia_block_amount
Skipping: 128
Skipping: /nvidia_block_amount


None of the machines have NVIDIA GPU's. Ideas?
13) Message boards : Number crunching : GPU just Stopped (Message 54978)
Posted 3 Jul 2012 by Brian Priebe
Post:
This morning, I woke up and to my suprise the GPU Milkyway@Home task's had hung on one workunit, elasping more than 3 hours.
Check your Windows System Event log for events from "Display" similar to:

Display driver amdkmdap stopped responding and has successfully recovered.


If you get one of these, any GPU calculation in progress is going to hang forever.

It's not necessarily the case that you have to reboot the machine. The hardware should have already recovered but the work unit is hung. Try suspending GPU calculation from the BOINC Manager "Activity" menu and then resume it. This always works for me.
14) Message boards : Number crunching : Why is Milky Way Taking all 4 CPU cores for one work unit? (Message 54484)
Posted 22 May 2012 by Brian Priebe
Post:
That's what happens when it's a multi-threaded app. When it runs it uses all available CPU's.
I believe there is a hard limit of 16 threads/WU. The multi-threaded WU's never use all of my 24-core machine.
15) Message boards : Number crunching : Astronomical In Progress WU Count (Message 52263)
Posted 6 Jan 2012 by Brian Priebe
Post:
Except I've never seen this before. What caused it? One is reminded of the WU gobblers who had thousands of WU's 'in progress' and never returned a single result.
16) Message boards : Number crunching : Astronomical In Progress WU Count (Message 52255)
Posted 6 Jan 2012 by Brian Priebe
Post:
This machine (http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=314881) is showing 250+ 'in progress' WU's going back several days when in fact it has only 13. I tried resetting the project but the problem has not gone away.

Any ideas?

EDIT: I detached from the project entirely and this solved the problem. But there are now 256 abandoned tasks listed that were apparently never present on the machine to begin with.
17) Message boards : Number crunching : HD7970 on the horizon .. (Message 52119)
Posted 30 Dec 2011 by Brian Priebe
Post:
oh my, no gpu usage while Windows has shut off your display
One has to hope that their review is in error. The HPC crowd would deluge AMD with complaints if it were true.
18) Message boards : Number crunching : ATI14 WU's 'Infinite' Run Time (Message 51966)
Posted 14 Dec 2011 by Brian Priebe
Post:
On two occasions recently I have had GPU WU's run to 100% completion then spend 9-15 hours post-completion doing God only knows what. I aborted both. One was
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=43258871. The other has since scrolled off.

Any ideas?
19) Message boards : Number crunching : Project has no tasks available (Message 51514)
Posted 26 Oct 2011 by Brian Priebe
Post:
We seem to be back to having no new tasks available again...
20) Message boards : Number crunching : Milkyway N-Body uses all BOINC time (Message 51300)
Posted 5 Oct 2011 by Brian Priebe
Post:
Odd. None of my "mt" WU's have higher than 1h20m expected run time. And even that was on a machine with DCF of 19.8 for MW.


Next 20

©2024 Astroinformatics Group