Posts by Jonathan Brier
log in
1) Message boards : Number crunching : IF U HAVE A BAD AMD/ATI CARD (Message 56114)
Posted 6 Nov 2012 by Jonathan Brier
You could also look at tuning down the frequency that Milkyway@home runs chunks of its simulation and checking to see if your cards are dust free... I mean really dust free.

My card was spitting errors and between the dusting out of the cards (twice) so they could cool properly again and moving the value to 80 in the preferences all is well now with my 5970 crossfire setup it appears to slice through the work units once again and is stable.
2) Message boards : Number crunching : Done WUs disappearing? (Message 54076)
Posted 20 Apr 2012 by Jonathan Brier
Even though the results disappear after 2 days are you tracking the failed/errored tasks in any way past that time?

I'm curious both in regards to: http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2856 I am also curious to what type of interfaces you use to monitor volunteer computing computational task submissions and what metrics you track (success rate, failure rate, validation but inconclusive, throughput time, etc).
3) Message boards : Application Code Discussion : Errors: Unhanded Exception Task 174285468 (Message 54051)
Posted 17 Apr 2012 by Jonathan Brier
[4] AMD ATI Radeon HD 5800 series (Cypress) (1024MB) driver: 1.4.1703

My host: http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=194804

Specifically an ATI 5970 running CCC 12.3

I understand there were some OpenCL bugs in previous drivers back to I think 11.9, but it seems a bit poor planning not to save the task with errors to debug and investigate further... unless that is done, but not publicly visible.

Next time I'll remember to capture the output.
4) Message boards : Application Code Discussion : Errors: Unhanded Exception Task 174285468 (Message 54012)
Posted 13 Apr 2012 by Jonathan Brier
The title really should read: Errors: Unhanded Exception Task 173626250
5) Message boards : Application Code Discussion : Errors: Unhanded Exception Task 174285468 (Message 54007)
Posted 13 Apr 2012 by Jonathan Brier
I've been experiencing random blue screens due to ATI driver crashes only when computing with Milkyway@home from what I can tell thus far. Still testing Collatz and Primegrid. The card processes many good work units thus this is not a common issue, but at unknown periods it will just crash. The errors very but for the most part it is a time out on recovery attempts.

I was combing through my error work units when I stumbled upon Task 174285468 which is the only one thus far that seems to have been able to run the debugger before my system BSOD:http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=173626250

I'm wondering if there is a bug in the code that may be causing my card driver to crash and the system to BSOD. Let me know if I can provide any more data or test runs to verify this.
6) Message boards : Number crunching : Crossfire and driver question (Message 51288)
Posted 3 Oct 2011 by Jonathan Brier
Hurrah with the 11.9 Catalyst drivers no crashes or hangs! This is impeccable timing. For the first time in half a year I am able to crunch on my machine for more than 5 hours and not have it crash. Its going on 15 hours straight and no crashes... will continue to crunch/test this new driver version. Hopefully it continues with its stability. :)
7) Message boards : Number crunching : What do you crunch on? (Message 51259)
Posted 30 Sep 2011 by Jonathan Brier
I built a new desktop in 2010 primarally for number crunching, but also for informational processing and programming. Its a beast of a machine for home use. :)

If you look at the graph you can see when I first turned it on for crunching.

http://boincstats.com/stats/boinc_user_graph.php?pr=bo&id=ce52cd1568c091014ffdef77dcd3fff2

Specs:
6 core AMD 1090T at 3.7Ghz
2x ATI 5970's in crossfire
16 Gb RAM
ssd boot w/ 4TB raid5 storage
Running most BOINC projects available.

Then there are my laptops one a Pentium M 2.0 Ghz, but that rarely is crunching as it is almost dead. The other laptop does Spinhenge and malariacontrol.net and is an AMD A8-3510 processor 4 core and only crunches while I someplace for a while.

8) Message boards : Number crunching : Crossfire and driver question (Message 51258)
Posted 30 Sep 2011 by Jonathan Brier
I too have been facing the BSOD with 11.4+ for the ATI catalyst driver. There is definitely some bug that needs to be addressed. Either by ATI or Milkyway. The latest driver 11.8 does well for longer time, but then it causes the driver to hang

Running Win 7 64bit 2x ati xfx 5970 in crossfire. > this computer is the one having issues: http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=194804

I keep on coming back hoping the problem is fixed but it is not yet... until then Miklyway does not get my GPU's computing for it. :(
9) Message boards : News : bypassing server set cache limits (Message 45763)
Posted 25 Jan 2011 by Jonathan Brier
I am glad to see this being posted in the news feed. Not everyone reads the forums often or at all. Providing announcements here is helpful for highlighting important issues such as this. Thank you for the update.
10) Message boards : Number crunching : Upload Pending - running out of work (Message 42825)
Posted 13 Oct 2010 by Jonathan Brier
Is anyone else experiencing periods of time where they cannot upload any of their Milkyway results and are stuck with Upload Pending for a long time (a few times an hour).

I have 2x ATI 5970's devouring WU and at times they are unable to unload the results to the server. Collatz can tear up the work-units and return no problem nor does any other project I am attached to have a problem uploading. If I suspend network activity then resume 2-6 work units are able to be returned then is stuck at pending again. It clears out a few times an hour and does not have issues then will repeat.

I'd like to donate more to Milkyway, but I cannot keep my cards fed with workunits...

Let me know what you think it might be or if you need more details.




11) Message boards : Number crunching : All GPU WU fail after ATI Catalyst 10.8 upgrade (Message 41770)
Posted 26 Aug 2010 by Jonathan Brier
Figured out the root of the problem... had 2 system changes and blamed the usual suspect when it was another. It is not the Catalyst update, but is dealing with monitor setup.

Situation where GPU Units function: Dual 5970 in non-crossfire or crossfire mode when 2 monitors are hooked up to one graphics card.

Situation where GPU Units do NOT Function: Dual 5970 in non-crossfire with 3 monitors hooked up. 2 on the 1st card and 1 on the second.

(crossfire does not work when using 2 cards and the 3 dvi connections for 3 monitors (card 1 x 2dvi, card 2 x 1 dvi hooked-up))

With a 2 monitor dvi setup on one card the WU work installing the 3rd monitor dvi on the second card dvi the work units do not progress.
12) Message boards : Number crunching : All GPU WU fail after ATI Catalyst 10.8 upgrade (Message 41756)
Posted 26 Aug 2010 by Jonathan Brier
I just updated to the new 10.8 Catalyst drivers and apps for my ATI 5970's and now all GPU apps are failing. I have currently suspended GPU WU downloads, but thought you might someone may be able to tell me what is causing the unhanded exception... bad install, incomparability, or what.

The behavior is the WU counts down the minutes to completion, but never stops nor does the progress indicator move from the initial 0.000%.

One of the WU that failed:
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=182384757

The collatz GPU WU fail in the same way as well.
13) Message boards : MilkyWay@home Science : Milkyway and other projects need help updateing info (Message 5632)
Posted 27 Oct 2008 by Jonathan Brier
How is this the wrong forum?

GridRepublic is working to become the unified public outreach for all projects and needs help filling in information for all of the projects on the site.
This includes science and related links. GridRepublic if you do not know is an account manager for managing BOINC projects.
14) Message boards : MilkyWay@home Science : Milkyway and other projects need help updateing info (Message 5613)
Posted 25 Oct 2008 by Jonathan Brier
GridRepublic

This account manager just released its community features which there already is an outline for content, but GridRepublic needs help of the BOINC community to help fill it out as they spend their time and effort continuing development and refinement of the site and beginning public outreach and partnerships to provide awareness for volunteer computing as a whole.

If you would like to work on outreach too the communities feature has an advocacy section where you can post and share discussions and files.

Also if you are a skilled web programmer, designer, or just interested in other ways you can help check out the developer section under the new community features added on October 24, 2008. They are a big step forward and we look forward to more improvements and strides in public awareness.




Main page · Your account · Message boards


Copyright © 2017 AstroInformatics Group