Points "earned"

[kofa]

Hello,

I'm folding on two machines (I joined a year ago, upgraded hardware two weeks ago; a password has been set on my account for a long time).
One runs Linux, is equipped with a Ryzen 5 5600X, 64 GB RAM and an NVidia 1060/6GB. Currently, its CPU is working on a WU for Project 16955, and reports 6607 points per day. The GPU is working on a WU of Project 17800, reporting 885052 points per day.
The laptop, running Windows, has an i5-10210U and 16 GB of RAM; its CPU is working on project 17231, and is reporting 32069 points per day.

How come the much more powerful Ryzen reports a much lower score? According to htop, the CPU usage of the CPU client is over 950%; the command-line of the process is:

Code: Select all

/var/lib/fahclient/cores/cores.foldingathome.org/lin/64bit-avx2-256/a8-0.0.12/Core_a8.fah/FahCore_a8 -dir 02 -suffix 01 -version 706 -lifeline 553738 -checkpoint 15 -np 11

Those (-np 11 and >950% CPU) seem consistent to me.

Thanks,
Kofa

[JimboPalmer]

I am unclear what client you are on, but in any case, F@H reserves one CPU thread per GPU. So you 12 thread Ryzen 5 5600X only tries to use 11 threads. (-np 11) However, some clients are very adverse to 'large' Prime numbers and numbers with large Prime factors. So 11 may well be shifted down to 9 to avoid 11 and 5 which are primes and only use 3 which is not 'large'.

So your Ryzen 5 5600X may not actually using all that many more threads than your i5-10210U. (9 vs 8) Now we are just down to processor design.

In general RAM is not a bottleneck for F@H.

If your laptop has a Passkey and your Desktop does not, you will get a Quick Return Bonus for the laptop.

Your GTX 1060 6gb is doing much better than mine on Windows 10, about 750,000 PPD

[gunnarre]

How come the much more powerful Ryzen reports a much lower score?

Enter your username here: https://apps.foldingathome.org/cpu
Does it say "1" in the "Got bonus" field in every line? If not, double-check that you put in the correct stats Passkey on the Ryzen system.

[Joe_H]

Besides what has been mentioned so far, any estimates you see reported by the client are possibly inaccurate. Accuracy is best after the client has continuously processed a WU for at least 3-5% since starting or resuming after a pause. Estimates are also more accurate for WUs from projects that client has processed in the past. The first time a WU os processed from a project the estimates are initially based on the Timeout value and assume the entire length of time will be used.

If you want a more accurate estimate of points, you can use this Bonus Points calculator - http://www.linuxforge.net/bonuscalc2.php - provided by a contributor of 3rd party apps. Use an average TPF based on actual elapsed times posted in the log, I usually use an average over 10%.

[kofa]

Both machines have the passkey set up. I think scores depend on the WU type.
For example, my laptop (the same device that printed '32069 points per day' for most of this morning) has started another WU, this time for Project 17231, and now prints 676 points per day, while the Ryzen now prints 16537 points per day.
I'm folding as 'IstvanAttilaKovacs', and yes, 'bonus' is 1 for all my CPUs:

Code: Select all

CPUid              IP Address       Team    Slot        OS      Last credit  Got bonus   Last returned         Project   Run    Clone   Gen
2A0C795E42F95895   xxx.xxx.xxx.115  243401  NVidia GPU  Linux   138,122      1           2021-03-31 11:11:07   13446     2420     30      0
280C795E42F95895   xxx.xxx.xxx.106  243401  x86         Linux     1,139.6    1           2021-04-08 06:41:21   17422        0   2125    208
6DA06E600A6DE1DA   xxx.xxx.xxx.106  243401  x86         Windows   9,442.6    1           2021-04-09 09:41:24   16815       26    918     35
290C795E42F95895   xxx.xxx.xxx.106  243401  NVidia GPU  Linux    47,826.2    1           2021-04-09 19:01:21   17800       55    152     57