| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: Africa Rainfall Project
Thread: Work Available |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 3315
|
|
| Author |
|
|
spRocket
Senior Cruncher Joined: Mar 25, 2020 Post Count: 277 Status: Offline Project Badges: 
|
I had checked on a whim to see what sort of results were showing up, and, whoa! ARP work units in flight? I took a quick look, and it was back to the same old transient HTTP errors, so my autoclicker had to come out of retirement. But hey, something more than MCM for a change!
Nice to see my Raspberry Pis get steady work as well - it's easy to overlook their contribution until they go missing. |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2171 Status: Offline Project Badges: 
|
On my device it's currently taking about 6 minutes, including retries, to download one ARP1-task.
Last night it has been downloading 64 ARP1-, 9 MCM1- and 60 OPN1-tasks, taking 7 hours (74 iterations of wcgresults -x). That's pretty quick, if you're taking the transient HTTP errors into consideration. 64 ARP1 × 6 minutes = more than 6 hours. I found that I needed to set LC_TIME=en_GB at the top of my crontab to have a nice date format to my liking. Compare: [Tue 24 Jan 2023 09:39:24 AM CET] Removing lockfile /tmp/wcgxmit.LOCK ... (that's without setting LC_TIME) and [Tue 24 Jan 2023 13:53:02 CET] Creating lockfile /tmp/wcgxmit.LOCK ... (that's with setting LC_TIME). Example for inclusion in a crontab: LC_TIME=en_GB |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12435 Status: Offline Project Badges: 
|
Adri
Good to see the date/time in the right order. Mike |
||
|
|
Unixchick
Veteran Cruncher Joined: Apr 16, 2020 Post Count: 993 Status: Offline Project Badges: 
|
This WU is 136 generation, but it took way longer than usual, even on my slow machine.
https://www.worldcommunitygrid.org/contribution/workunit/251502922 Could it possibly have a smaller time step? |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2171 Status: Offline Project Badges:
|
That's hard to tell, once the task has left your system, without any other information, Unixchick.
----------------------------------------Normally, when the task is running, first find out in which slot your task is running, e.g. on the commandline: $ wcgresults -NO1 — let's say it was the third task, then your slot would be '14' — or use the BOINC Manager, click on the name of the task in the "Tasks" view, then click the Properties button, then you should see its Directory with the slotnumber, e.g. "slots/14". Then you can go to the designated 'slots'-directory (you should be a member of the group 'boinc'), e.g. ~boinc/slots/14, where your ARP1-task is running. Once you're there, you can carry out a search: $ grep time_step= namelist.input The 'normal' time-step is 36, the smaller one is 24 (you already knew that, of course). Of course, you can also watch the percentage increasing in the "Tasks" view; Mike.Gibson has explained before how this corresponds to the time-step.  Adri [Edit 1 times, last edit by adriverhoef at Jan 24, 2023 7:18:25 PM] |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12435 Status: Offline Project Badges:
|
Unixchick
It is simpler to look at the progress column in BOINC Manager. If it goes up in increments of 0.020%/0.021% (actually 0.0208333%) then that is the standard 36 second timestep. 36 seconds is 1% of an hour and there are 48 hours in the 2-day workunit A 24 second timestep progresses in steps of 0.01388888% (0.013/0.014%). But you can only see that while crunching. 20 hours is not a particularly long time. I take 21-24 hours normally. A 24 second timestep would take about 50% longer than normal. Mike |
||
|
|
Unixchick
Veteran Cruncher Joined: Apr 16, 2020 Post Count: 993 Status: Offline Project Badges:
|
A normal ARP takes 15-16 hours normally. This one took 21 hours.
So probably not a smaller time step. I wish I had caught it while it was still on my system. Thanks for the reminder of the info. I need to make myself a file of all the useful info from this thread so I don't have to hunt for it. |
||
|
|
MJH333
Senior Cruncher England Joined: Apr 3, 2021 Post Count: 268 Status: Offline Project Badges: 
|
A normal ARP takes 15-16 hours normally. This one took 21 hours. Unixchick,My ARP1s are also taking longer than normal. I think, in my case, it’s because most of my threads are running ARP. Mike advises a maximum of 50% threads on ARP and I would normally abide by that limit, but I’m not getting enough MCM at the moment to do so. Cheers, Mark |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12435 Status: Offline Project Badges:
|
Mark
I have no problems with supply at the moment. My 2 8-thread machines are set to cache 12 ARP1, 8 MCM1 & 8 OPN1 and to run 4 ARP1, 2 MCM1 & 2 OPN1. HST1 & SCC1 are also permitted but not available so as to get some when they return. Mike |
||
|
|
MJH333
Senior Cruncher England Joined: Apr 3, 2021 Post Count: 268 Status: Offline Project Badges:
|
I have no problems with supply at the moment. Thanks Mike.I think my supply issue is the result of running TN-Grid as well as WCG, not running a cache and not using an app_config.xml file. I’m not really worried about it! Cheers, Mark |
||
|
|
|