| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Retired Forums
Forum: UD Windows Agent Support [Read Only]
Thread: FAAH, checkpoints and when to snooze |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 5
|
|
| Author |
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18665 Status: Offline Project Badges: 
|
I normally have the WCG agent running whenevery my machine is running, which is normally any time I am not in the process of commuting to/from work. Before shutting down, I've always tried to wait to go into snooze right after a checkpoint. With HPF, I believed that to be when the three graphs reached the right side of their windows. Then, when I started back up, the WCG agent would not have to "redo" any work since the last checkpoint. I would see the percent complete on the application panel start back at the same percent (to one decimal place) value. With FAAH, I am still trying to figure out when I can presume a checkpoint has occurred. My impression at this point is that it is when the "Best Docking Energy" window's blue line restarts at the left. I realized I'm looking at minimal CPU cycles here but, like anything with grid computing, multiplied across all users, it could make a difference when you are trying to maximize your contribution and not waste cycles. I've read that there will eventually be more frequent checkpoints in FAAH but this would still be relevant then albeit less so. Any suggestions on how to best visually confirm when a checkpoint occurs?
---------------------------------------- |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi keithhenry,
On HPF, if you went to the Application View, every time the progress went up a few tenths of a percent, it had just checkpointed and computed the progress bar. This is not true for FAAH. Rick Alther says that FAAH only checkpoints when the green line (maybe blue on your screen) reaches the right edge. I do not know how to tell when it checkpoints if it does not find a good docking site and never starts drawing a green line. We need more check points. mycrofth |
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: 
|
I do not know how to tell when it checkpoints if it does not find a good docking site and never starts drawing a green line. We need more check points. I had one workunit like that, that never had any docking at all. I was watching UDMonitor at the time, which still managed to do some checkpoint saves. It knew when the ligand finished, but it wasn't visible on the graphic because it didn't dock. mycrofth The UDMonitor log should show when a save was just finished...and the flask icon would then change from yellow or red (if it has been awhile) to blue. Every workunit is different. Some save only 4 minutes apart. I had one that took 41 minutes between saves. I also had occasion to shift slots to send in some results. When I went back, it restarted at the last invisible checkpoint and showed a reasonable percentage of completion, based on where I left off. In other words, if you are using UDMonitor, it is still possible to watch and shut down right after a checkpoint.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 4 times, last edit by retsof at Dec 4, 2005 2:53:14 PM] |
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18665 Status: Offline Project Badges:
|
Rick Alther says that FAAH only checkpoints when the green line (maybe blue on your screen) reaches the right edge. I do not know how to tell when it checkpoints if it does not find a good docking site and never starts drawing a green line. We need more check points. mycrofth Oops, it's green for me too. Just happened to have a WU at the time of my post that had never started drawing a green line (there was no red either) even after being about 1/3rd done. What I have noticed is typically the case is that I seem to "lose" 2-3% when restarting the agent as goes goes back to the last checkpoint. Apparently, that can be much more siginificant but I've just not caught it happening to me. I'll just wait for the "tune up" where the checkpoints become more frequent. FWIW, it also seems that, for a given amount of runtime, FAAH returns slightly fewer points that HPF does. Not sure how much the difference is but it's enough that it appears to be noticeable looking at my statistics history.  |
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18665 Status: Offline Project Badges:
|
The UDMonitor log should show when a save was just finished...and the flask icon would then change from yellow or red (if it has been awhile) to blue. Every workunit is different. Some save only 4 minutes apart. I had one that took 41 minutes between saves. .... In other words, if you are using UDMonitor, it is still possible to watch and shut down right after a checkpoint. I have considered using UDMON but have only seen it being a benefit when there are server outages that leave you in a paused state waiting to return a result. However, this may provide a second reason. |
||
|
|
|