Despite having opted out half way through the beta because my GPU is too slow and it’s playing hob with the rest of the tasks I find I’ve downloaded 18 of the production tasks on one of my machines.

Looks like each task will take just under 3 hours so I’ll let them run for now and see just how it affects my setup.

If it gets too bad I’ll switch off nvidia, that should stop them :-)

For reference, I am running 3 at a time on a 1660 Super and 3 on a 970 without any problems. Each WU finishes within 3-5 min with little deviation. By running 3 at a time, I reach roughly ~95% sustained load with less jumps in GPU util. Successfully rocked that config during several beta runs. Your card should easily be able to run a higher load than mine. Cheers!

I confer. Hardly ever saw more than 30 jobs packed into the beta WUs. Now one of the first OPNG WUs packed 74! jobs and finished in 297sec on a 1660S. That's just ridiculous compared to CPU equivalent expected runtimes that seem to always pack 2 jobs each (?) and take about ~1.5 hrs on a Zen2 Ryzen CPU.

Hmm, beyond coincidence. The first two OPNG production tasks to run (OPNG_0000008_00094 & OPNG_0000008_00107) have given computation error after 02:27:56 exactly.

Does that rule out timing problems with the GPU card but not, for example, memory limitations? (Only 1gb on the card).

I’ll leave it running overnight but if the third task ends after the same time I’ll cut my losses and bow out (again).

Yep, exactly the same and now I can get to the stderr file I can see it’s timing out after 36 jobs.

Sorry, no work from this end - not powerful enough.