17919 - Did one slip through the cracks?

[BobWilliams757]

This project did hit beta and was there for a few days, then was released to FAH.

It's giving very high PPD returns.

https://folding.lar.systems/gpu_ppd/b ... _rtx_3090

About twice (or more) than average on a 3090. The 3070 about normal PPD.

My 1660 Super got crazy high PPD. 12.78m on about 82 watts.

I'm suspecting something is a bit strange on this project. I've only picked up the one, and the others were already on LAR when I looked.

[BobWilliams757]

RTX 4080 hits 133M PPD.

[kiore]

RTX 4080 hits 133M PPD.

Are you actually getting those credits or is this a projection? Really looks like something amiss here.

[BobWilliams757]

All the ones on LAR are people other than me, so I can't speak for that.

But I got the 12.78M on my 1660 Super, and was credited as such, almost 222K points. Less than a half hour, 15 second TPF.

Going back on the Generation from mine, it appears everyone is getting huge numbers or base credit.


I could have contacted the researcher through here, but I didn't know if that was within forum rules.

[Joe_H]

How many steps were reported for the WUs you saw from 17919? There have been bugs in the past where the WS would change the number of steps from the initial setting the WUs were benchmarked on.

The other possibility is that the size in atoms for this project is towards middle sized. Possibly processing on high end GPUs is using the bonus points faster up the exponential curve than usual.

[BobWilliams757]

Joe,

250,000 steps. 103,824 atoms.


If we see something like this is the future, what's the preferred way to send up a flare? PM or email the researcher, or one of the admins, Beta members? I just assumed that the researchers don't want emails from regular folders on any frequent basis, and wasn't sure if making comment or note in the Announcements/Consortium section would draw the researchers notice, same for PM as I don't know if most of them check out the forum frequently.