Advanced search

Message boards : Graphics cards (GPUs) : Task Checkpointing Every 90 seconds

Author Message
Dayle Diamond
Send message
Joined: 5 Dec 12
Posts: 84
Credit: 1,663,883,415
RAC: 1
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 57038 - Posted: 30 Jun 2021 | 6:39:05 UTC

e3s333_e1s689p0f967-ADRIA_New_KIXcMyb_HIP_AdaptiveBandit-1-2-RND5072

The CPU time since last checkpoint isn't going over the 90 second mark.
It's supposed to save a maximum of every 600 seconds.

My last upload for an Adria New Adaptive Bandit task was half a gigabyte, and the task took over a day to compute.

Saving anything like this to my SSD every 90 seconds for days is going to wear it out.

Is this happening to everybody else, and how do I get the task to obey BOINC limits?

Dayle Diamond
Send message
Joined: 5 Dec 12
Posts: 84
Credit: 1,663,883,415
RAC: 1
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 57336 - Posted: 20 Sep 2021 | 4:02:39 UTC - in response to Message 57038.

Still happening, all these months later.

e1s718_I907-ADRIA_AdB_KIXCMYB_HIP-0-2-RND8712_2

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 57341 - Posted: 20 Sep 2021 | 8:47:31 UTC - in response to Message 57336.

The task is checkpointing as required by the app to work. But it should not be saving every 90 elapsed seconds.

The total size of the result is large.

gdf

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1617
Credit: 8,101,644,351
RAC: 18,354,676
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 57343 - Posted: 20 Sep 2021 | 9:32:24 UTC - in response to Message 57341.

The total size of the result is large.

I have no problem with a large result file, provided you keep an eye on the maximum size allowed in the task specification, and ensure there's always plenty of headroom.

There's nothing more frustrating than completing a large task, but having the result rejected by the server because the size has grown beyond an outdated limit.

Post to thread

Message boards : Graphics cards (GPUs) : Task Checkpointing Every 90 seconds

//