Well, based on the number of Nbody WUs that are waiting to be sent, it looks like the WUs that need re-validation were kicked back to the surface! I'm going to let things run. If you wind up with problems getting work, please let me know and I will figure something out.

If we can just let this run then I think all those "stuck" WUs will finally be processed, although it will take a while for people to work through 11M of them...

Can someone check on the leftover WUs that still need to be validated, and look at a couple things for me? I'd like to know what the following values are for these WUs:

1) server_state
2) outcome
3) client_state
4) validate_state
5) need_validate

This should be found in a combination of the "result" and the "WU" information. I'm not sure what things look like on the client. If you send some links to a few of these WUs that might be the most helpful since I can directly compare these stuck WUs with the freely-flowing WUs.

Then, hopefully, I can manually force these WUs to generate new validation WUs, but I can do it ~10k WUs at a time or some other reasonable number so that we don't blow up the server.

What do you think of all these tasks previous to April 20?

https://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=412567034
https://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=413057030
https://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=413674439
https://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=414290323

I've got other 3000 like them

Tom tried various methods to reduce the huge backlog of N-body tasks that seemed to result from the recent server problems; the last (and apparently successful) attempt was referred to in the Nbody WU Flush News post, and it is the reason for those "Didn't need" retry tasks.

That's the ""Why?" -- the real question is "What happens next?" -- rather than typing yet another magnum opus :-) I'll refer you to my earlier reply on this topic in this very thread...

I have a much smaller number of tasks stuck in this state, so I have been monitoring this and will continue to do so -- if these tasks are still stuck in that state by the end of April, Tom will need to tell us what is going to happen next...

Cheers - Al.

288larsson, would you please mind running again this Einstein test with the last drivers to check whether there's has been any improvement?

Another question, was the card in quiet or uber mode during the test?

Thanks in advance.

Hi uber mode during the test.
I can run the test after primegrid challenge end Jan 18

Hello,

I was wondering if you could test out your R9-290x with running three Einstein BRP5 tasks (0.33 utilization factor) and post what the run time is per task? If you could run a batch of GPU tasks without any other CPU tasks running that would be great.

Thanks,

Jeroen

Hi 8390sec Catalyst 13.11betav8