Inconsistent PPD on 1080

[TonyStewart14]

I just came back to folding after a long hiatus and noticed a big difference in the first two WUs, both COVID-related. The first was project 13418 which gave a whopping 5.5 million PPD on a GTX 1080 with only 40% utilization! The second was 11759, which was more expected at 750K PPD at close to 100% utilization. I'm sure that PPD is not the priority compared to getting WUs out, but is this a common thing? Note that I am running the 1080 on a XG Station Pro, but I don't think the external GPU is causing it since the second WU appears to be fine.

[HaloJones]

P13418 has some interesting sub-units that can be unbelievably productive for one's personal numbers.

[bruce]

Supposedly the points were set high to make up for the high failure rates of some Runs in P134xx.

[TonyStewart14]

That makes sense, thanks!

[bruce]

You may not be so lucky when the next batch of production projects are launched (soon).

[JohnChodera]

The 134xx projects have larger RUN-to-RUN variations than we had expected. We're still working on minimizing this effect.

We had an incident today that resulted in some short WUs with very large PPD being released into running projects---see here for more info:
viewtopic.php?f=19&t=35867#p340260

~ John Chodera // MSKCC

[LazyDev]

Wonder if this is why I've noticed my RTX 2080 Super doing over 20M PPD when it was working on the moonshot, earlier.

[MeeLee]

Wonder if this is why I've noticed my RTX 2080 Super doing over 20M PPD when it was working on the moonshot, earlier.

That's more than 200 CPUs folding