| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: Africa Rainfall Project
Thread: Loss of processing time |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 10
|
|
| Author |
|
|
richardcyb
Cruncher Joined: May 2, 2007 Post Count: 10 Status: Offline Project Badges: 
|
Has anybody noticed that on shut down of your computer, say, at the ennd of the day something like 12% or 4 hours of computing time is lost when you turn on your computer the next day?
|
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges: 
|
This project makes 1 checkpoint every 12.5% of progress. Shut it down before it does so and you lose all work since the last checkpoint.
----------------------------------------It is how it works.  - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz [Edit 1 times, last edit by Falconet at Apr 17, 2020 10:12:19 AM] |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12594 Status: Offline Project Badges: 
|
Exactly as Falconet says. The way round it is to Hibernate your machine instead of Shutting Down. Then you won't lose your processing time.
Mike |
||
|
|
richardcyb
Cruncher Joined: May 2, 2007 Post Count: 10 Status: Offline Project Badges:
|
Thanks for that. I'll hibernate in future. It does not seem to apply to other projects.
Rich |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges: 
|
Thanks for that. I'll hibernate in future. It does not seem to apply to other projects. Rich It does apply tother projects, it is just that the checkpoints are so far apart, about 3 hours or so for a job which takes 24 hours for ARP versus times which can be measured in minutes for other projects which have much shorter work units. You do notice when you miss 3 hours of work versus missing 10 minutes of work. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12594 Status: Offline Project Badges:
|
MIP is another where you can lose a significant amount of time, although those have variable checkpoints.
Mike |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Uh, every 1/8 of total process?
Can modify this checkpoint to every 5% of process (like 2 hours)? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The jobs feed off 6 hour interval meteorological data and have 8 of those intervals packaged, that's why the 12.5%. The program is actually trying to validate if the the highly granular data can be simulated accurately into the future. Saving is convenient at those points, as the output can be used as input to re-launch. Mid-run would require writing the whole in-memory model of up tp 1GB to storage and load it again on resume, that's what your hibernating does once a day. You will have noticed that's quite intense on the disk and likely to be noticed by users.
----------------------------------------[Edit 1 times, last edit by Former Member at May 19, 2020 10:42:15 AM] |
||
|
|
bidon
Cruncher Joined: Dec 9, 2005 Post Count: 26 Status: Offline Project Badges: 
|
In my case, hibernation don't work with ARP.
First time, i use directly the "hibernate" button. The ARP task doesn't restart. Second time, i put the boinc client into snooze before hibernate. It goes to previous check point after few minutes... |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As the Minimum System Requirements table in Help and the Device Profiles footnote for ARP1 "t
• Please review the system requirements before opting to participate in this project" , recommended setting is to activate Leave Application In Memory (while suspended), aka for short as LAIM. The description in the client is slightly different "Leave non-GPU tasks in memory while suspended". So, what you did unloads ARP1. More so, since BOINC will be suspended just before the OS is hibernated, you might be creating some limbo state causing it to not be able to restart properly. Suggest to active LAIM and try again. |
||
|
|
|