| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Discovering Dengue Drugs - Together
Thread: Completion of DDDT? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 15
|
|
| Author |
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: 
|
Sekerob,
----------------------------------------I am sorry for having confused you by making reference to the 60-second save interval that you mentioned and that I am using too. I did it only to make clear that in my PC no checkpointing opportunity is missed because of this setting. Beside this I have made no mention of checkpoints occurring in DDDT at fixed time intervals or not, therefore there is no need to elaborate about this. And I agree that this is not the case. What I said and what I am sure of for having observed many DDDT WUs running is that DDDT WUs checkpoint at "round" percentages, i.e. when the percentage is precisely 1.000, 2.000, 3.000, etc... and consequently there are 100 possible checkpoints in every DDDT WU, at least for the WUs that have been distributed until today. Normally there should be 99 because there is obviously no checkpoint at 100.000 %, but one checkpoint is also taken at 0.000 % after the initialisation period, just before the percentage starts to increase. If you are still unsure of this I can send you a stdoutdae.txt file where you will be able to count those 100 checkpoint messages per DDDT WU. And then I am back to my initial question, is there really one docking per checkpoint and 100 dockings in every DDDT WU? I have nothing against it but I have the feeling that it is much when compared to other WCG projects. I hope I have been more clear this time. Jean. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Can someone deconstruct a work unit to find out how many dockings each one contains? I thought the techs determined how much to pack into a WU? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I thought the techs determined how much to pack into a WU? They do. I will send a query, rather than try guessing.  Lawrence |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
Greetings all,
There are a few things that determine the check pointing in a work unit. The program tries to checkpoint after every docking. This is the main thing that determines it, but it will check your BOINC settings to make sure that if you've check pointed more frequent then your settings, it will skip a check point. This means that if you set your BOINC settings to check point at most every 10 minutes and a docking finishes quicker than that, it will skip that check point and continue onto the next docking. Now, onto the number of dockings per work unit. There are currently 100 dockings in the DDDT work units. This is also set by the researchers on how many dockings they need for each calculation they want to run. The docking number can be dynamic as you have noticed with FAAH. It is based off how difficult a work unit is and if we need to chop up the work unit into smaller sets of dockings. For example, if a work unit is estimated to take 20 hours and we have the work unit generator set to create a max of 10 hour work units, then the number of dockings would be set to 50 for each work unit. I hope this helps, -Uplinger |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
I'm glad to see that the rule of checkpoint and attempt/docking/step relation is generally still valid (GC was the sole exception so far i think). That said it might be worth considering to introduce a 5 minute wallclock check to save, to cut on disk i/o. Who's bothered with an chance average crunching loss of 2.5 minutes on a faster machine.
----------------------------------------Separately, it does seem like where i previously understood the science would ask the daemon if it was okay to save, it does only check that at start up. Playing around with the saving value which usually i have at 5 or 10 minutes, changing it to 60 seconds let it continue as if the 5 minute minimum was still in place. A bug in the 5.10.30 and .38 client maybe? Hence my false impression like there were 20-22 in the present run. thanks
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|