With a 3060 Ti, I had this project a couple times, and as it's a cancer project, points are usually on the very low side, but this time, it's abysmal. An ETA of 18 hours with a 3060 Ti, for a PPD of 220 000 or less. The card uses 54-60w instead of it's typical power. Normally, that project should yield between 1M and 1.5M PPD, but this time, it's even less.
Project 17605 (Run 315, Clone 1, Gen 69)
Project 17605 : Extremely long ETA
Moderators: Site Moderators, FAHC Science Team
-
pyrocyborg
- Posts: 36
- Joined: Wed Sep 28, 2022 1:45 am
- Hardware configuration: 3060 12GB, 3060 Ti, 3070, 3080, 3090, 6700 XT, 6800 XT, 6900 XT
-
sukritsingh
- Scientist
- Posts: 141
- Joined: Sat Mar 14, 2020 11:53 pm
Re: Project 17605 : Extremely long ETA
Interesting! This is a project where as more gens are added there's more energy/bias in the system due to the sampling method used. It makes sense that there would be this slowdown after a while (low gen WUs were performing fine it seems). This is really interesting from a technical standpoint since nobody's ever been able to run these samplign methods at this scale to actually see this slow down in ETA. I'll turn off the project now so more bad WUs aren't sent out, but thanks for uncovering this!