| 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: 3268
|
|
| Author |
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12398 Status: Offline Project Badges: 
|
Mark
A time step of 18 secs would take almost twice as long as for a 36 secs. unit. which ties in with your forecast of 25 hours. It would seem that 18 secs is the third alternative after 36 & 24. 36 secs is 1% of an hour, 24 would be 2/3 of 1% and 18 would be 0.5%. Mike |
||
|
|
MJH333
Senior Cruncher England Joined: Apr 3, 2021 Post Count: 267 Status: Offline Project Badges: 
|
The average CPU time on ARP1 tasks I've completed on this system recently is about 15.75 hours. The 102 task is still running, expected to take about 25 hours total. Just to confirm that the 102 task took 24.5 hours, about 55% more than average for my system. Cheers, Mark |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12398 Status: Offline Project Badges:
|
The last time there was any movement of Ultras was mid April when they only moved 1 generation. Now they and other extremes seem to be attempting multiple moves.
Recently, the only automated output seem to have been re-sends. Does this mean they have now enabled automatic regeneration? Has Dr Jurasica turned into Dr Who? Mike |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 721 Status: Offline Project Badges:
|
Not really much point in having work available when we can't upload anything
---------------------------------------- Currently being moderated under false pretences |
||
|
|
MJH333
Senior Cruncher England Joined: Apr 3, 2021 Post Count: 267 Status: Offline Project Badges:
|
Adri,
I've picked up a couple more extremes: ARP1_0034390_116, and ARP1_0033952_118. Both are time_step 36. Incidentally, the 102 task is now Pending Verification, so I wonder if that is going to fail.  Cheers, Mark |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2167 Status: Offline Project Badges: 
|
Hi Mark, did you have a good night's sleep? I went to bed
 with a large upload queue on one of my systems, thinking there'd be nothing I could do at the moment but to wait and rely on 'WCGtools', to finally get up in the morning with an empty upload queue  and a big thank you to 'wcgresults -x' .Hey, that's nice work, Mark, with the 'url=http' lay-out etc. that you used!  I've taken a peek at the URLs, recorded them in my spreadsheet and noted that, by the looks of it, there are two identical Sent times: ARP1_0034390_116_4 MSWin 10 In Progress 2023-05-08T07:15:23 2023-05-09T19:15:23 So I gather you must have picked them up on one device at the same time  (one and one makes two). Give the device a slap on the keyboard or a pet on the mouse from me for such nice work.  My system that I refer to as 'm2' seemed to have picked up the Extreme task ARP1_0033791_108_0 at 2023-05-08T01:37:11 this morning. It's nearly finished (in two hours).  Adri |
||
|
|
Aperture_Science_Innovators
Advanced Cruncher United States Joined: Jul 6, 2009 Post Count: 139 Status: Offline Project Badges: 
|
Got a question about a WU one of my devices is chewing on that maybe you pros know more about. This system normally runs ARP WUs in 15-16 hours each, However, it has one that it's been running so far for about 32 hours, and still expects another nearly 9 hours left:
----------------------------------------Application: Africa Rainfall Project 7.32 Name: ARP1_0033793_102 State: Running Received: 5/6/2023 11:20:18 PM Report deadline: 5/8/2023 11:20:17 AM Estimated computation size: 215,958 GFLOPs CPU time: 1d 08:24:30 CPU time since checkpoint: 01:48:57 Elapsed time: 1d 08:30:27 Estimated time remaining: 08:40:12 Fraction done: 78.944% Virtual memory size: 815.77 MB Working set size: 715.77 MB Directory: slots/125 Process ID: 72972 Progress rate: 2.520% per hour Executable: wcgrid_arp1_wrf_7.32_x86_64-pc-linux-gnu Here's a link to the WU. Any idea why it would take so long? Given that it's due back in 4 hours and won't finish by then, any point in letting it keep running?  [Edit 2 times, last edit by Aperture_Science_Innovators at May 8, 2023 1:53:08 PM] |
||
|
|
MJH333
Senior Cruncher England Joined: Apr 3, 2021 Post Count: 267 Status: Offline Project Badges:
|
Any idea why it would take so long? Given that it's due back in 4 hours and won't finish by then, any point in letting it keep running? This seems to be one of the grid cells that have proved tricky to compute and have fallen behind the main pack. I suspect that the WU is taking longer because the "time_step" has been reduced in an attempt to get it to compute successfully. You can check by opening the "namelist.input" file in the relevant slot directory and searching for "time_step". The normal time_step is 36. The researchers have been using 24 on problematic grid cells. And recently we have had an 18 too. If yours is 24 or 18, that would explain the additional CPU time. I would leave it to run to its conclusion. Cheers, Mark |
||
|
|
MJH333
Senior Cruncher England Joined: Apr 3, 2021 Post Count: 267 Status: Offline Project Badges:
|
Adri,
Hey, that's nice work, Mark, with the 'url=http' lay-out etc. that you used! I was copying from a pro!  I gather you must have picked them up on one device Yes, my main PC. I've been lucky with interesting WUs this week Cheers, Mark |
||
|
|
Aperture_Science_Innovators
Advanced Cruncher United States Joined: Jul 6, 2009 Post Count: 139 Status: Offline Project Badges:
|
Any idea why it would take so long? Given that it's due back in 4 hours and won't finish by then, any point in letting it keep running? This seems to be one of the grid cells that have proved tricky to compute and have fallen behind the main pack. I suspect that the WU is taking longer because the "time_step" has been reduced in an attempt to get it to compute successfully. You can check by opening the "namelist.input" file in the relevant slot directory and searching for "time_step". The normal time_step is 36. The researchers have been using 24 on problematic grid cells. And recently we have had an 18 too. If yours is 24 or 18, that would explain the additional CPU time. I would leave it to run to its conclusion. Cheers, Mark Thanks for the explanation, Mark. Don't have access to the system to check now, but next time I see this happen I will look at it. I wonder why they drop the time_step (& thus make the WU take 2x as long to complete) _and_ also give it such a short deadline. Seems like a poor combination.... |
||
|
|
|