| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: The latest batch occasionally fail half way through |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 21
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Computation errors (from other similar projects) are a normal part of finding the end points in chemical computations with GPU's. Less often encountered with CPU computations because they are returned so much slower.
----------------------------------------However the server wishes to address them is usually determined by their choice of programming language, All, let me say again, ALL computations involving Genetic/Chemical/Hormonal computations will involve long term/involved guesswork. That's what they need US for. Just seems that HCC(1) has not figured how to grant credit when the returned data crunched falls outside the expected. JIMO, YMMV Edit: Haven't been 'here' that long but have 'been' a donor for a long time. If your heart is in it for a cure, just hang in there. I get a lot of points, but I'd do this for NONE if it came to that :) [Edit 1 times, last edit by Former Member at Nov 20, 2012 8:11:13 PM] |
||
|
|
David Autumns
Ace Cruncher UK Joined: Nov 16, 2004 Post Count: 11062 Status: Offline Project Badges: 
|
After another day of carefully watching .... I have spent way too much of my life testing and trialling telecoms hardware and software.... it is failing at the beginning of the GPU starting to crunch through the second work unit
----------------------------------------I think this is the same error condition as the also very few work units that fail at the beginning of the first The only problem with failing on the second is that the perfectly crunched first one never makes it home It's a trade off - you have just doubled the risk of failure with 2 images in one WU. As the GPU pauses between the images, without an app_info.xml file, I don't think the efficiency of the project improves with 2 images in one WU. It just makes for half as many WU to count and collate in the database which I am sure helps Kevin and Crew. Dave  [Edit 1 times, last edit by David Autumns at Nov 21, 2012 8:46:22 AM] |
||
|
|
coolstream
Senior Cruncher SCOTLAND Joined: Nov 8, 2005 Post Count: 475 Status: Offline Project Badges: 
|
I think you are spot on with what you are saying.
----------------------------------------I know the techs are busy, but I hope they are at least reading this and that they will be able to provide some sort of feedback at a later date.  Crunching in memory of my Mum PEGGY, cousin ROPPA and Aunt AUDREY. |
||
|
|
David Autumns
Ace Cruncher UK Joined: Nov 16, 2004 Post Count: 11062 Status: Offline Project Badges:
|
 I haven't had one of these failures since 1pm yesterday Don't tell above a dozen Thanks go out to the back room crew  Dave |
||
|
|
David Autumns
Ace Cruncher UK Joined: Nov 16, 2004 Post Count: 11062 Status: Offline Project Badges:
|
 spoke too soon My point score yesterday shows what reliable GPU crunching could provide to the project. If they are liable to fail could we have just 1 Image per Work Unit for now That way they fail after a few seconds instead of half way through and we can sprint onto the next good one Dave |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
I had all sorts of problems (from errors to BSODs) when using an app_info to run more than one WU. But now that I run only one work unit per card, I never get failures on two HD 7770s on two different PCs.
I think that simply demonstrates why WCG does not officially support app_info files, and I doubt that the techs will be spending any time trying to "fix" it. They have enough problems just keeping up with the work returned by the good cards that are not having problems. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
same here jim no errors at all
|
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
I had all sorts of problems (from errors to BSODs) when using an app_info to run more than one WU. But now that I run only one work unit per card, I never get failures on two HD 7770s on two different PCs. I think that simply demonstrates why WCG does not officially support app_info files, and I doubt that the techs will be spending any time trying to "fix" it. They have enough problems just keeping up with the work returned by the good cards that are not having problems. There is nothing the techs can"fix" to solve the issues you or anyone else is having. There are who knows how many crunchers running an app_info without any issues. If you had problems then there is something on your system causing it. I'd be glad to help you troubleshoot it if you'd like.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
  |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
Thanks, but there is nothing to fix. And I don't think you quite get the point. The app_info may work fine on some cards in some systems, but not others. I don't think there is a viable solution; it would be a custom "fix" in each case, which simply means that you would be overcoming the shortcoming of app_info by various work-arounds in each case.
----------------------------------------EDIT: And my HD 7770s are fairly well-matched to a single core of my CPUs, so there is not much more to be gained without impacting other projects. I think it is more worthwhile on the higher-level cards, if you are willing to devote more cores to it. It makes more sense for me if they devise a way to prevent the CPU from trying to run more than one work unit at once; then I can more usefully run two work units on my cards. [Edit 2 times, last edit by Jim1348 at Nov 30, 2012 3:06:23 PM] |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
Thanks, but there is nothing to fix. And I don't think you quite get the point. The app_info may work fine on some cards in some systems, but not others. I don't think there is a viable solution; it would be a custom "fix" in each case, which simply means that you would be overcoming the shortcoming of app_info by various work-arounds in each case. I'm not proposing any "workarounds" because there simply aren't any. Give me your OS, type of CPU. type and amount of mobo memory and driver version then we can try to customize an app_info that will work for you. ]EDIT: And my HD 7770s are fairly well-matched to a single core of my CPUs, so there is not much more to be gained without impacting other projects. I think it is more worthwhile on the higher-level cards, if you are willing to devote more cores to it. Actually there is quite a bit more to be gained if you can run multiple GPU tasks on your card. It makes more sense for me if they devise a way to prevent the CPU from trying to run more than one work unit at once; then I can more usefully run two work units on my cards. That is what setting the correct parameters in an app_info file will do.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
----------------------------------------[Edit 1 times, last edit by nanoprobe at Nov 30, 2012 8:00:47 PM] |
||
|
|
|