| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: GPU Overload |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 4
|
|
| Author |
|
|
Bryn Mawr
Senior Cruncher Joined: Dec 26, 2018 Post Count: 344 Status: Offline Project Badges: 
|
I finally decided it was time to dip my toe back into the water and try another GPU task as I remember hearing that the time limit has been increased and my very slow GPU (gtx710) might have a chance of completing a task.
I mistakenly assumed that Boinc would follow the usual process of sending a single task to make sure the host could process the new type of task. Sadly it sent me 28 GPU tasks in a single download. In an hour or so, when the first task finishes, I could well be aborting 27 of them. Oh well, I live and learn :-) I’m just about to turn off the tap - even if it runs ok it will take me about 3 days minimum to clear this lot and I definitely cannot cope with any more. |
||
|
|
spRocket
Senior Cruncher Joined: Mar 25, 2020 Post Count: 274 Status: Offline Project Badges:
|
Why abort them, instead of just letting them run? Does the card overheat? {edit: also, there's the whole slowdown issue} They're worth the credit even when you have a slower GPU. You'll get hundreds of BOINC credits per hour instead of thousands, but that's still a good deal more than CPU crunching alone.
----------------------------------------I'd love to have a backlog long enough to keep the GPU going 24x7.  [Edit 1 times, last edit by spRocket at Nov 3, 2021 8:07:43 PM] |
||
|
|
Bryn Mawr
Senior Cruncher Joined: Dec 26, 2018 Post Count: 344 Status: Offline Project Badges:
|
Why abort them, instead of just letting them run? Does the card overheat? {edit: also, there's the whole slowdown issue} They're worth the credit even when you have a slower GPU. You'll get hundreds of BOINC credits per hour instead of thousands, but that's still a good deal more than CPU crunching alone. I'd love to have a backlog long enough to keep the GPU going 24x7. If the first task had run for 2 hours and the crashed having taken more than the time limit (which is what happened last time I tried) then I would have aborted the rest to save wasting my computer time and not hold up processing the tasks for the project. As it is the task completed after 2:45 so I’ll happily let the rest run. |
||
|
|
|