Points Drop

[Bob_6591]

Hello

Over the last few days I noticed a very large drop in points. This surprised me as up until now my points have been fairly consistent. I understand that different projects/work units have different point values. Is this normal to have an extended period of very low value work units? All three computers are set to accept whatever F@H throws at me.

Thanks.

Date Day League Rolling 24h PPD Change 24h PPD 24h WU WU CH
2024-1-14 Majors 1,425,988 609,512 22 3
2024-1-13 Minors 804,102 -280,007 19 -3
2024-1-12 Majors 1,084,109 195,020 22 9
2024-1-11 Minors 889,089 171,743 13 5
2024-1-10 Minors 717,346 -1,834,530 8 -6
2024-1-9 Majors 2,551,876 -581,756 14 -1
2024-1-8 Majors 3,133,632 236,762 15 -1
2024-1-7 Majors 2,896,870 -378,511 16 -1
2024-1-6 Majors 3,275,381 -304,489 17 2
2024-1-5 Majors 3,579,870 -2,317,101 15 -11

[BobWilliams757]

There are a number of things which can impact points, but the luck of the assignment server alone can be a big factor at times. If you frequently use the system for other things that has impact as well. Looking at logs will tell you if you have had any errors that caused folding to restart from a checkpoint, and that costs time and as such lowers PPD.

If you use HFM it can give you a better idea of the assignment server lottery, and if you check your logs on occasion to look for errors that can help as well.

Sometimes folding for a specific cause will increase consistency in points, simply because you will do more of the same projects in greater numbers rather than have a lot of different projects.

[Bob_6591]

“There are a number of things which can impact points, but the luck of the assignment server alone can be a big factor at times.”

Big is the right word. Here is an example of what I did versus a teammate. I did three times the work units for about 1/5th the credit. This is a strange point system.

Day Points WUs
01.14.24 1,338,748 25 - me
01.14.24 5,689,200 8 - teammate

“If you frequently use the system for other things that has impact as well. “

I do not think this is it. Been doing the same thing for the last two months.

“Looking at logs will tell you if you have had any errors that caused folding to restart from a checkpoint, and that costs time and as such lowers PPD.”

Not sure where the logs are located. Would this be an issue since the work units are getting completed and https://folding.extremeoverclocking.com/ has them listed on their website?

“If you use HFM it can give you a better idea of the assignment server lottery, and if you check your logs on occasion to look for errors that can help as well.”

Never heard of HFM. Did a quick internet search to find out. It is not something listed in the Mint software manager. Not sure how to install it.

“Sometimes folding for a specific cause will increase consistency in points, simply because you will do more of the same projects in greater numbers rather than have a lot of different projects.”

I am going to give this a try. Maybe it will break the curse.

The graphics card I am using is a RTX 3060 Low Hash Rate – anti-mining card. I do not think this is a problem, but who knows. The good news, I should be getting a new RTX 4060 in the next two days. It will replace the 3060.

I would like to leave this open for a few more days to see how things shake out so that I can provide an update.

Thanks for your help.

[calxalot]

I believe HFM is currently Windows-only.

viewtopic.php?t=9903

[bikeaddict]

The current log is at /var/lib/fahclient/log.txt and the log archive is under /var/lib/fahclient/logs/ on Fedora Linux, not sure about Mint. You can run commands like these to look for problems:

Code: Select all

grep -i exception /var/lib/fahclient/log.txt
grep -i fail /var/lib/fahclient/log.txt
grep -i exception /var/lib/fahclient/logs/*
grep -i fail /var/lib/fahclient/logs/*

[BobWilliams757]

It seems like your work unit count is high for the GPU, but that might just be the freak luck of the assignment server and your card species. It might be worth checking out your performance vs the average for those with the same GPU on https://folding.lar.systems/

If you use the drop down for "research projects" you can search by project for the averages they get submitted to them, which should be at least somewhat close to what you get unless you have something strange going on. You can get the project details from the logs, as well as time specifics and/or errors. Unless you are getting short projects that don't produce many points something seems out of whack.... high work unit count suggests speed but your PPD isn't very high for the GPU you are using.

If the EOC website shows the completed units, it only shows those completed and credited AFAIK, and not any that failed. But you can have errors that slow things down and still complete as well. This would be obvious in the logs (any certain % numbers that repeat, and the appropriate restart info) and in time stamps.


And unless things have improved with PPD since you've posted, I'd look for any system logs, OS logs or such things with recent changes after Jan 9th or so, since that is when your points took a dive.

[Bob_6591]

Thanks for all the ideas.

bob@bob-desktop:~$ grep -i exception /var/lib/fahclient/log.txt
grep: /var/lib/fahclient/log.txt: No such file or directory
bob@bob-desktop:~$ grep -i fail /var/lib/fahclient/log.txt
grep: /var/lib/fahclient/log.txt: No such file or directory
bob@bob-desktop:~$ grep -i exception /var/lib/fahclient/logs/*
grep: /var/lib/fahclient/logs/*: No such file or directory
bob@bob-desktop:~$ grep -i fail /var/lib/fahclient/logs/*
grep: /var/lib/fahclient/logs/*: No such file or directory

I did find the fah-client folder. There are three folders (cores, credits, work) and three files (client.db, config.xml, gpus). A search of the directory revealed three files with log in the name (2x73…loG… .json, logfile_01.txt & science.log). The second and third files appear to represent the current work unit as the date changed from 1/15 when I looked at it yesterday to 1/16 this morning. The json file had a creation date of Dec 23 so it would not be applicable.

I have been looking at the folding.lar.systems. It is one of the reasons I bought a RTX 4060 (115 watts). The 4060 gets me more points (1 million +) for a lot less power (RTX 3070 = 170 watts). I wanted an efficient card for the summer. It will also save about $48 a year in electricity.

54 GeForce RTX 3060 Lite Hash Rate GA106
2,652,701
2,800,569
2,603,271

29 GeForce RTX 4060 AD107
3,946,586
4,131,125
3,545,434
https://folding.lar.systems/gpu_ppd/overall_ranks

The numbers from folding.lar.systems do suggest that the 3060 card is rather poor at folding. In general performance is not that much slower than the 4060 or my 2070S.

Yesterday the numbers for the 3060 were: 2024-1-15 Minors 544,993. Another odd thing is that Folding@Home In The Dark v8 Web Client: BETA has been saying the Current Est PPD is 3 million or so a day for over a week. No idea where they get their number from.

It looks like I am getting some really bad work assignments. It is a bit frustrating. The plan now is to wait for the 4060 to arrive, it is 190 miles away! It should be here tomorrow. The 3060 will be banished to the gaming PC (non-folding) and the 2070S will get paired with the i3.

Thanks again for all your help.

[bollix47]

Try the following and report their output (first line is for v7 and the second is for v8):

Code: Select all

ls -l /var/lib/fahclient
ls -l /var/lib/fah-client

If you don't get any output from both of those commands then fahclient was not installed in the default directory and you will have to tell us where you installed the software.

[Bob_6591]

bob@bob-desktop:~$ ls -l /var/lib/fah-client
total 376
-rw-r--r-- 1 fah-client fah-client 81920 Jan 17 20:11 client.db
lrwxrwxrwx 1 root root 25 Nov 7 07:38 config.xml -> /etc/fahclient/config.xml
drwxr-xr-x 5 fah-client fah-client 4096 Nov 23 04:22 cores
drwxr-xr-x 2 fah-client fah-client 61440 Jan 17 11:05 credits
-rw-r--r-- 1 fah-client fah-client 222452 Jan 12 09:46 gpus.json
drwxr-xr-x 3 fah-client fah-client 4096 Jan 17 17:20 work

Sorry this is late, I just finished replacing the 3060 with the new 4060. Also had a few things to do today.

Tomorrow I will be replacing the GTX 1050 with a 2070S. Between the two cards I hope to get back up to 6 Million PPD.

[Bob_6591]

I decided to send an email to Gregory Bowman, PhD (The Director of Folding@home). Here is part of my email: “...I was unable to find a technical reason for the drop in points. It appears as if the server is assigning very low value work units to my two computers. I would like to maximize my contribution to the project....”

Professor Bowman responded with: “… I'll check into it.”

Thanks for your help.
Cheers.

[BobWilliams757]

So you're still getting low PPD with the 4060 and the 2070 Super?

[Bob_6591]

Yes.

Date.....Rolling 24h PPD .....24h WU ...Expected Points Millions.....Graphics Card Folding
2024-1-22.....1,086,268.....22......…………….............7….....RTX 4060, RTX 2070S
2024-1-21.....1,048,770.....27........……………...........7….....RTX 4060, RTX 2070S
2024-1-20.....1,644,575.....16.......……………............7….....RTX 4060, RTX 2070S
2024-1-19.....1,234,317.....20.....………….….............7….....RTX 4060, RTX 2070S
2024-1-18…....811,893.....16.......……….……...........4.…....RTX 4060, GTX 1050
2024-1-17…...893,668…....7.....………….….............4….....RTX 4060, GTX 1050
2024-1-16.......921,556.…..10.....…………….............4….....RTX 4060, GTX 1050
2024-1-15…....544,993.....15.........…………….........2.5…...RTX 3060 LHR, GTX 1050
2024-1-14.....1,425,988.....22........……………..........2.5…...RTX 3060 LHR, GTX 1050
2024-1-13…....804,102.....19.........…………….........2.5…...RTX 3060 LHR, GTX 1050

[bollix47]

What is your folding name?

With it we would be able to see your folding status at:

https://apps.foldingathome.org/

It is looking like your passkey is missing:

https://apps.foldingathome.org/getpasskey

Please type the following in a terminal and post the results here using the code tags:

Code: Select all

cat /etc/fah-client/config.xml

** sorry for the missing dash above

[PaulTV]

This file may contain an existing pass key and may contain other security sensitive info which you should not post here. Please simplky check if the passkey is in the file. Example snippet of my config, with the actual values changed:

Code: Select all

  <!-- User Information -->
  <passkey v='<some large hex number that I really don't wanna post here>'/>
  <team v='<the team number for which I fold>'/>
  <user v='<my folding user name>'/>

[Bob_6591]

What is your folding name?

With it we would be able to see your folding status at:

https://apps.foldingathome.org/

It is looking like your passkey is missing:

https://apps.foldingathome.org/getpasskey

Please type the following in a terminal and post the results here using the code tags:

Code: Select all

cat /etc/fahclient/config.xml

PHEFS28

bob@bob-desktop:~$ cat /etc/fahclient/config.xml
cat: /etc/fahclient/config.xml: No such file or directory

Is fahclient version 7.x and fah-client ver 8.x?

The config.xml is under /var/lib/fah-client
It shows "link (broken)". When I try to open it says to move it to trash.

Did a search for "config.xml". The five other files that have config.xml in the name are related to other programs.

The folding at home client "https://app.foldingathome.org/0/settings" shows that I have a passkey. I did send a request for my passkey. They gave me a new one. I tried opening the client on another computer. It froze with a message saying "connecting". Tomorrow I will try again.

I entered a passkey when I first started folding. Is it possible that the server lost it?