Something wrong with the WU's I'm getting?

[Samfisher]

Starts off strong on a Core 21 WU, shows 300K+ on my OC'd 970 as usual.  ETA is 6 hours 45 mins.  Leave for 12 hour job, come back, has 3 hours left, ETA now 3 hours.  PPD at 200k.  Restart PC in attempt to fix it.  Never goes to 300K again, stuck at 132k PPD. 6 hours later, ETA 2 hours.  5 hours later ETA is 1 hour.  2 hours after that ETA is 1 hour.  What the hell is going on?

 

TLDR : 6 hour 45 minutes WU has taken over 26 hours to complete, and PPD has shrunk almost 60%..  Wtf?

[tobben]

You are sure you are on the same wu? it does seem like it, but it's nice to be sure.

You keep mentioning ETA though, what about the progress on the actual wu?

Check the log.

 

Have you checked the load on your gpu? made sure it's stable? Have you checked 

the temperature to make sure it isn't throttling?

 

You could try resetting it to stock settings, Folding can be pretty sensitive when it

comes to overclocks on gpus. But it is rather strange, if you had a sour overclock

it would usually have failed the wu by now. Atleast from my experience. 

[Samfisher]

You are sure you are on the same wu? it does seem like it, but it's nice to be sure.

You keep mentioning ETA though, what about the progress on the actual wu?

Check the log.

 

Have you checked the load on your gpu? made sure it's stable? Have you checked 

the temperature to make sure it isn't throttling?

 

You could try resetting it to stock settings, Folding can be pretty sensitive when it

comes to overclocks on gpus. But it is rather strange, if you had a sour overclock

it would usually have failed the wu by now. Atleast from my experience. 

 

Yes it's the same WU cos I set it to Finish, which would end when the current WU is done.

 

The WU does increase by % on logs.

 

Load is fine, 97%-99% all the time.  Isn't throttling, never goes above 80c, and I'm on a no-throttle custom BIOS.

 

OC and non OC gives the same issue ): I've folded on this GPU for awhile before, almost 7m points to LTT, but just this one WU has been totally nuts.  It still says 45 mins to completion right now.

[tobben]

Yes it's the same WU cos I set it to Finish, which would end when the current WU is done.

 

The WU does increase by % on logs.

 

Load is fine, 97%-99% all the time.  Isn't throttling, never goes above 80c, and I'm on a no-throttle custom BIOS.

 

OC and non OC gives the same issue ): I've folded on this GPU for awhile before, almost 7m points to LTT, but just this one WU has been totally nuts.  It still says 45 mins to completion right now.

 

Right, seems like it might just be an oddball wu then. Atleast if it doesn't happen again. 

I guess you have to decide for yourself, whether you are going to stick with it and try

to finish it, having already spent 30 hours or so on it , or if you want to delete it and

try a different one.

 

Do you know how to delete a borked wu? 

[Samfisher]

Right, seems like it might just be an oddball wu then. Atleast if it doesn't happen again. 

I guess you have to decide for yourself, whether you are going to stick with it and try

to finish it, having already spent 30 hours or so on it , or if you want to delete it and

try a different one.

 

Do you know how to delete a borked wu? 

Yes I do, but I'll just wait it out I guess.  What's 45 mins (if it says 45 mins) after 30 hours

[tobben]

Will be interesting to see if the wu goes through or not, keep us updated. 

[Samfisher]

Will be interesting to see if the wu goes through or not, keep us updated. 

It actually went through! Woo, now for the next one!

[petem]

Starts off strong on a Core 21 WU, shows 300K+ on my OC'd 970 as usual.  ETA is 6 hours 45 mins.  Leave for 12 hour job, come back, has 3 hours left, ETA now 3 hours.  PPD at 200k.  Restart PC in attempt to fix it.  Never goes to 300K again, stuck at 132k PPD. 6 hours later, ETA 2 hours.  5 hours later ETA is 1 hour.  2 hours after that ETA is 1 hour.  What the hell is going on?

 

TLDR : 6 hour 45 minutes WU has taken over 26 hours to complete, and PPD has shrunk almost 60%..  Wtf?

 

Sounds like the "slow down" bug.  I had a similar problem on a core 21 project 9704 wu.  At 38% it went from a 2:07 m:s TPF to almost 2 hours.  I didn't notice it until the morning (gotta sleep sometime :rolleyes:) at which point I stopped that slot, and went online for answers.  No luck there, but when I came back and restarted the slot (it jumpbed back about 5% to the previous checkpoint), it went back to 2:07.

 

I posted a query on the official forum, and got a response informing me "The development team is aware of the problem and a possible fix should be in the next version of the core which is coming "soon"."  It doesn't usually break the WU, but can really cause it to lag.  I lucked out and just happened to get it back on track.  I probably waited about 1/2 an hour while I was browsing the forums.  He said 30 seconds is all it takes.

 

Apparently it is uncommon enough that they don't feel it is necessary to pull the project and fix the problem.  So ... guess we have to be WU babysitters :unsure: 

 

I suggested they add a feature to alert someone or to automatically try an automatic reset if there is a drastic change in performance, but they are hoping that fixing the core will save them from having to update the client.