| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: GPU Work has resumed |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 95
|
|
| Author |
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
We have released a new validator that should dramatically reduce the issue that we saw over the weekend with a very large number of invalid workunits and as a result we have resume distributing work.
This version will do a different type of check when the best energy in a set of runs is still positive. It will now check this across 2 or more returned results and make sure that both results find similar outcomes. Previously if the set of runs has a positive least value for the runs, the result was automatically marked invalid. Note that due to the results being returned by different hosts for the same workunit have slight variances in their energy computations, this check cannot be an exact check but it is instead checking if values are within a narrow range. As a result there are sometimes where you will see one of your results get marked invalid even though the device has a history of running well and unfortunately this is the nature of these type of "fuzzy" validators. |
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 1878 Status: Recently Active Project Badges:
|
Thanks for that. Are you going to increase the output of OPNG WU's for a while, as you mentioned before?
----------------------------------------  |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
I'm going to let it run overnight at this speed and confirm it is working as intended first. If it is, then I'll increase the speed in the morning.
|
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 1878 Status: Recently Active Project Badges:
|
I'm going to let it run overnight at this speed and confirm it is working as intended first. If it is, then I'll increase the speed in the morning. Very good. Looking forward to more OPNG WU's, even if it's only temporary. |
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 1878 Status: Recently Active Project Badges:
|
@knreed
----------------------------------------Look at this one as an example of the 0/0 Cpu time/ Elapsed time, that's been happening on the results list, since the move to the new website. I have plenty of them on CPU tasks, and now one for OPNG. Mine is the W8.1 with the Cpu time/ Elapsed time. They do validate, but the Cpu time/ Elapsed time is 0/0. https://www.worldcommunitygrid.org/contribution/workunit/817074363 [Edit 1 times, last edit by Grumpy Swede at Sep 22, 2021 9:12:22 PM] |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
Thanks for pointing that out. If you look at the API response that backs that page (https://www.worldcommunitygrid.org/api/v3/ms/workunit/817074363) you will see that when the cpu/elapsed time gets small enough it is being returned in SI notation instead of just as a decimal and that isn't being converted properly for display.
That should be an easy fix. |
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 1878 Status: Recently Active Project Badges:
|
Still one invalid OPNG. The W8.1 machine is mine:
----------------------------------------https://www.worldcommunitygrid.org/contribution/workunit/817193521 |
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 1878 Status: Recently Active Project Badges:
|
Thanks for pointing that out. If you look at the API response that backs that page (https://www.worldcommunitygrid.org/api/v3/ms/workunit/817074363) you will see that when the cpu/elapsed time gets small enough it is being returned in SI notation instead of just as a decimal and that isn't being converted properly for display. That should be an easy fix. Well, small enough, I don't know. It happened on CPU tasks that took over 4 hours too (on my very slow Laptop), as well as my faster cruncher, that took about 1.5 hours to crunch OPN1. Edit: And the 0/0 issue does not happen immediately after reporting, but after a couple of minutes, the right crunch times turns into 0/0. Edit2: And it does not happen all the time, but maybe 1 in 20-40 tasks, or even less.than that. No matter crunching times. [Edit 3 times, last edit by Grumpy Swede at Sep 22, 2021 9:28:53 PM] |
||
|
|
Richard Haselgrove
Senior Cruncher United Kingdom Joined: Feb 19, 2021 Post Count: 360 Status: Offline Project Badges: 
|
Good to return home and see a bundle of OPNG tasks in the list. But less good to see that the majority of them are _2 and _3 replications.
Some research to be done in the morning, methinks. |
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 1878 Status: Recently Active Project Badges:
|
They seems to be replications from some of those that went invalid on the 19th.
----------------------------------------Edit: Not good so far though. I still get invalids. [Edit 2 times, last edit by Grumpy Swede at Sep 22, 2021 9:44:55 PM] |
||
|
|
|