Force RMSE error, p17717

[Bastiaan_NL]

I just noticed the following error on one of my systems and I hope sharing it here is useful:

Code: Select all

21:06:40:WU02:FS00:0x22:ERROR:Force RMSE error of 6.1443 with threshold of 5
21:06:40:WU02:FS00:0x22:Saving result file ..\logfile_01.txt
21:06:40:WU02:FS00:0x22:Saving result file science.log
21:06:40:WU02:FS00:0x22:Saving result file state.xml
21:06:41:WU02:FS00:0x22:Folding@home Core Shutdown: BAD_WORK_UNIT
21:06:41:WARNING:WU02:FS00:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
21:06:41:WU02:FS00:Sending unit results: id:02 state:SEND error:FAULTY project:17717 run:24 clone:0 gen:44 core:0x22 unit:0x000000000000002c0000453500000018
21:06:41:WU02:FS00:Uploading 6.75MiB to 128.174.73.74

After looking for the specific unit on the WU Status page, I found out that I'm not the only one.

I already found the following answer:

The RMSE error message is from the part of the FAHCore that validates that the WU is coherent and is intended to stop processing if there has been a gross error before you waste a lot of time just to have the WU discarded later. The threshold of 5 is supposed to allow for floating point truncation errors and other factors where the calculations are "close enough" vs. "Wrong"

Obviously 6.7 is only a little too much error, but you have to set a threshold somewhere.

[Neil-B]

Subject project and others mentioned are still beta projects .. posting to beta forum will let researcher know .. posting here doesn't help beta progress

[Bastiaan_NL]

Thanks for forwarding it Neil-B, I saw your post over there.
I would have posted there but I don't have access.

[Neil-B]

This tbh is why beta flag should only be used by beta team .. errors need to be caught early and the researchers need to know .. might be worth you signing up for the beta team if you are running the flag