| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: [Fixed...ATM] Running out of GPU WUs fast!!! |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 42
|
|
| Author |
|
|
coolstream
Senior Cruncher SCOTLAND Joined: Nov 8, 2005 Post Count: 475 Status: Offline Project Badges: 
|
It looks like I have run out of GPU WUs again. All machines set to do HCC GPU work only are dry. ...another good reason to have an apps_info.xml (looking for GPU and CPU tasks) installed to pick up the slack and run CPU tasks during the 'dry' spells. As soon as a new GPU task is grabbed, it starts to run automatically and then reverts to CPU again when no more GPU are available.   Crunching in memory of my Mum PEGGY, cousin ROPPA and Aunt AUDREY. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sometimes I run out of both GPU & CPU on my quad core I5 It hasn't been massive. Say approximately 3 percent of the time I run out of both. Approximately 10 percent I run out of one or the other. I have been fiddling with the settings, so I don't know how much this means. This is my working computer so I have gotten in the habit of "watching the clock," so to speak. I turn it off - most nights.
----------------------------------------[Edit 2 times, last edit by Former Member at Dec 9, 2012 5:06:30 PM] |
||
|
|
mclaver
Veteran Cruncher Joined: Dec 19, 2005 Post Count: 566 Status: Offline Project Badges: 
|
It looks like I am getting all I want now. I am going to create a cache for my GPU only machines with app_info so I do not run out. I usually do not keep a cache.
----------------------------------------   |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
It looks like I am getting all I want now. I am going to create a cache for my GPU only machines with app_info so I do not run out. I usually do not keep a cache. 400 tasks is the max you can get for the GPU no matter how high you set your cache.
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.
  |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
Ok - we will have to think on this. The addition of the Linux GPU versions on Friday has resulted in those app versions repeatedly clogging up the shared memory array. This is what is causing the issue with people not getting work. Let me see what we can do.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I did like how quickly the linux GPU tasks went from Beta to acceptance as a regular task. But does that last comment mean they may go back to beta?
|
||
|
|
coolstream
Senior Cruncher SCOTLAND Joined: Nov 8, 2005 Post Count: 475 Status: Offline Project Badges:
|
All in all, the news of another 6 months of WUs (at present rates) plus an explanation of the 'no work available' messages have indeed been good news.
----------------------------------------I envisage Santa being very busy delivering ATI cards this Xmas Crunching in memory of my Mum PEGGY, cousin ROPPA and Aunt AUDREY. |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
I have made some changes that should improve the ability to fetch work for the project. I've also increased the allowed GPU cache up to 600 and you can fetch 15 workunits per scheduler request.
It is hard to see on the server side if people are able to stay supplied with work. Can you please post if you have been able to maintain a steady supply of GPU work? thanks, Kevin |
||
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges: 
|
452 In progress ATM
----------------------------------------Edit to update: 600 WU's In progress ATM  Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006  [Edit 1 times, last edit by branjo at Dec 11, 2012 4:44:17 PM] |
||
|
|
mclaver
Veteran Cruncher Joined: Dec 19, 2005 Post Count: 566 Status: Offline Project Badges:
|
All in all, the news of another 6 months of WUs (at present rates) plus an explanation of the 'no work available' messages have indeed been good news. I envisage Santa being very busy delivering ATI cards this Xmas I missed the notice of 6 months more of HCC. That would be good news becasue with GPU processing, I am now doing over 3,000,000 ppd and have moved up from an overall rank of 38 to 33, and should hit 31 in one more week! All of this is due to GPU processing. What does the N/A under days left mean under HCC on Sekerob's charts, this is what I usually use to approximate the days left. http://i137.photobucket.com/albums/q210/Sekerob/WCGYearsPi1Project.png |
||
|
|
|