A host with a CUDA device checkpointing every second for 25-30 minutes!
6 Feb 2010 12:12:27 UTC
Topic 194761
(moderation:
Hi, after a few error's were brought to my attention, by MarkJ, BTW, thanx again, I looked at some result's and see that it's been checkpointing like crazy, but does validate.
Can anyone explain this phenomenum?
Using BOINC 6.10.18.
WU
and this one
Looked at 10 of them, all the same.
A host with a CUDA device checkpointing every second for 25-30 m
)
Hi!
If you go to http://einstein.phys.uwm.edu/prefs.php?subset=global, what is your setting for: "Tasks checkpoint to disk at most every", and did you overwrite this in the local preferences in BOINC Manager??
CU
HB
RE: Hi! If you go to
)
Hi Bikeman, was just going to post and saw your post, looked at a validated task ABP2, they are called?
Noticed this checkpointing again look at this WU!
Yes, write to disk=0, will change that, should make some difference. :)
And very certain, overwritten local, too.
Had thought of that a little late.
[ADDED]Takes up (a lot)of time and destroys my disks :)
The usual setting for
)
The usual setting for checkpointing to disk would be like every 60 seconds. If you do it more ofter, it's degrading performance and it's annoying to have the extra disk activity.
CU
HBE