Folding Forum

I'm seeing 30-40% PPD reduction for the 1922x CPU projects compared to other CPU projects.

Dell Inspiron 3880
i5-10400 (6 cores enabled)
12GB RAM (DDR4 3200, 1x8+1x4, non-matching dual channel)
80GB hard drive (WD800AAJS)
Ubuntu 24.04 LTS

19229 is taking 18 hours to complete for 63K PPD. 19227 and 19228 are taking 14 hours to complete for 80-95K PPD

All other projects running on this machine get close to 100-120K PPD.

Is it normal for the 1922x projects to be this slow? Is it possible they need to be benchmarked again?

19229 looks to be getting dumped repeatedly every day. That doesn't seem good for the science.
https://api.foldingathome.org/project/1 ... ne/0/gen/0

Yes, PPD varies between projects.
I swings of up to 1m PPD on my CPUs

p19229 is indeed below typical PPD on my CPU ...

And don't look at Run 0 / Clone 0 / Gen 0 reports, there's a bug in the work servers that counts all dumped WUs in the project as Run 0 / Clone 0 / Gen 0.

muziqaz wrote: ↑Tue Feb 03, 2026 5:30 pm Yes, PPD varies between projects.
I swings of up to 1m PPD on my CPUs

I realize PPD varies between projects. It's the percentage that's concerning. 30-40% seems a bit high.

To see if the difference is machine related, I increased cores from 3 to 4 on another Dell machine to pull 192xx work.

Dell Optiplex 7020
i5-4590 (4 cores enabled)
8GB RAM (DDR3 1333, 2x4, matching dual channel)
240GB SATA SSD
Windows 10 Pro

1922x work may be better served being sent to machines with >=8 cores instead of >=4 cores.

AgrFan wrote: ↑Tue Feb 03, 2026 8:41 pm

muziqaz wrote: ↑Tue Feb 03, 2026 5:30 pm Yes, PPD varies between projects.
I swings of up to 1m PPD on my CPUs

I realize PPD varies between projects. It's the percentage that's concerning. 30-40% seems a bit high.

To see if the difference is machine related, I increased cores from 3 to 4 on another Dell machine.

Dell Optiplex 7020 with a i5-4590 (4 cores enabled), 8GB RAM (matching dual channel), 240GB SATA SSD, Windows 10 Pro.

1922x work may be better served being sent to machines with >=8 cores instead of >=4 cores.

That will not solve the problem