No work assigments...

[SandyG]

I've had ongoing issues with no work units for a while, assignment servers vary. I can leave the system trying to get a new WU for a log time to no avail. What does work every time tho is to reboot the computer.

At a guess I'd say the assignment server connection is stalling out & then not clearing the previous attempt from its cache so never proceeds to get a response; note that this is intermittent but once stalled a reboot is all that seems to fix it.

Does the V8 beta behave differently in this regard & would be worth trying?

I have had various success with rebooting, doesn't always work. I think you can accomplish the same by stopping folding via the WebUI or other means and restarting. What I believe we see is that after a while the back off algorithm to prevent flooding gets really long, and by then, coincidentally, work units are available. Reboot and Restart shows the immediate attempt for the WU, but it's 50/50. I sat and did it last night about 10 times on an idle 3070 on my windows machine to no avail. Watched it hit the severs, only to have errors of no WU avail.

Oh well, still waiting for on for a while on a 4090.

[SandyG]

I've got BOINC on my boxes and it will nicely cede the card to F@H when there is work for Folding. I run Einstein (an astrophysics project that sifts through radio telescope data looking for various things) as the work units are not time dependent. Just one way to ensure my boxes are always doing some kind of science.

I have run BOINC for a bit on other projects when Seti@home folded... pun intended

Always found it difficult to find good projects. I loved Seti@home because it was kinda a waste of CPU but still super interesting, and never felt that from my work people would profit. I have always been torn with FAH but for the greater good of making advancements to help folks, I'm OK with it, more so then often I appear to be

Let's see where this all goes. At some point will start a new thread in the general section to discuss.

Sandy

[toTOW]

Sorry for the delay, but it's been hard to find someone alive in the labs ... I guess it's not the best timing to have such issues ...

But I finally found someone who has been able to add some work !

On a side note, there are 3 ways to create more WUs on FAH :

• create a new project ... the process takes a lot of time and requires a researcher
• add more runs to an existing project ... it's quicker since it doesn't require additional testing, but it still requires a researcher
• add more clones to an existing project. The process is automatic and the server handles the work, but adding clones doesn't help in some studies where the length of simulation (number of gens) is more interesting than adding more parallel data. More clones also requires more time to analyse the results ...

Anyway, say hello to project 18725 : https://stats.foldingathome.org/project/18725 from server at 131.239.113.97 (fahwork01.psivant.com). This is a rather new server, so let us know if you notice something unusual.

Also, the project is rather small (78k atoms) but it has been set to assign only to bigger GPUs to help with the shortage, so let us know if you see abnormal PPDs on your GPUs. I got one WU on a 4070 and a 4070 Ti and numbers seemed in range ...

[SandyG]

Sorry for the delay, but it's been hard to find someone alive in the labs ... I guess it's not the best timing to have such issues ...

But I finally found someone who has been able to add some work !

On a side note, there are 3 ways to create more WUs on FAH :

• create a new project ... the process takes a lot of time and requires a researcher
• add more runs to an existing project ... it's quicker since it doesn't require additional testing, but it still requires a researcher
• add more clones to an existing project. The process is automatic and the server handles the work, but adding clones doesn't help in some studies where the length of simulation (number of gens) is more interesting than adding more parallel data. More clones also requires more time to analyse the results ...

Anyway, say hello to project 18725 : https://stats.foldingathome.org/project/18725 from server at 131.239.113.97 (fahwork01.psivant.com). This is a rather new server, so let us know if you notice something unusual.

Also, the project is rather small (78k atoms) but it has been set to assign only to bigger GPUs to help with the shortage, so let us know if you see abnormal PPDs on your GPUs. I got one WU on a 4070 and a 4070 Ti and numbers seemed in range ...

Awesome work, and a huge Thank You for digging in to this. I have one of the 18725 WU running as I type this that just started running. Will see if it's seems reasonable (from the best that I could tell).

Again thanks for getting someone to look at this.

Sandy

[SandyG]

toTow -

Just looked at today's history and saw a couple of WU's completed on the 18725 Project, here are some of the numbers if that helps -

Code: Select all

RTX3090 NVidia GPU 	Linux 	898,600 	1 	2023-09-05 23:40:27 	18725 	0.074 Days
RTX3090 NVidia GPU 	Linux 	904,400 	1 	2023-09-05 23:45:38 	18725 	0.073 Days
RTX3070 NVidia GPU 	Windows 788,200 	1 	2023-09-05 23:45:37 	18725 	0.097 Days
RTX4090 NVidia GPU 	Linux 	1,158,200	1 	2023-09-06 00:45:30 	18725 	0.045 Days 	

HTH

Sandy

[Lazvon]

Calculated PPD (easiest to report) of four cards running 18725 right now:

4090: 16.4M (that one, xx424 IIRC, does 24M)
4080: 13.3M
4070Ti: 8.6M
3090: 7.7M

Thanks for the details!

[Andre_Ti]

Calculated PPD (easiest to report) of four cards running 18725 right now:

4090: 16.4M (that one, xx424 IIRC, does 24M)
4080: 13.3M
4070Ti: 8.6M
3090: 7.7M

It would be more correct to add at what power limit and GPU frequency these results were obtained.
For example, in project 18725, my 4090s at a frequency of 2900-2950 MHz and a consumption of 250W have an average PPD of 23.5M.

[SandyG]

Calculated PPD (easiest to report) of four cards running 18725 right now:

4090: 16.4M (that one, xx424 IIRC, does 24M)
4080: 13.3M
4070Ti: 8.6M
3090: 7.7M

It would be more correct to add at what power limit and GPU frequency these results were obtained.
For example, in project 18725, my 4090s at a frequency of 2900-2950 MHz and a consumption of 250W have an average PPD of 23.5M.

Are your numbers correct? I see on your history 10x less, on the couple that show up here -

https://apps.foldingathome.org/cpu

Also I do see maybe 60 watts lower processing on the 4090's and similar 250/260watts. Not sure about the other cars. All good as another heat wave coming here.

Looking quiet from the standpoint of errors and lack of WU's which is a very good thing!!!

Sandy

[Andre_Ti]

Are your numbers correct? I see on your history 10x less

I think they are true. For example, the last entry with 1,115,600 points on project 18725.

[SandyG]

Are your numbers correct? I see on your history 10x less

I think they are true. For example, the last entry with 1,115,600 points on project 18725.

I read the M as Millions not thousands, I'm thinking electronics when I see m - 1,000,000 and k as 1,000. Didn't really think about the case of the letter and in some cases folks use M for 1000. All good seeing much of the same

Sandy

[Andre_Ti]

I read the M as Millions not thousands, I'm thinking electronics when I see m - 1,000,000 and k as 1,000.

At the address you indicated, not PPD, but the number of points received and there is no hint of the abbreviations M, K, unit of measurement 1. To calculate the PPD, use - http://www.linuxforge.net/bonuscalc2.php.

[Lazvon]

M for millions… that PPD is out of HFM when those work units were running…

I do not tweak the cards, whatever out of the box they are set for, no voltage changes or the like, I do set an 80 degree limit, but on the 4xxx series never seen anything over 60-65 when I have looked. I have seen 3090 get in the 70s.

I won’t bother giving any more numbers.

[Lazvon]

Oh, and as the first line of the post said, those were estimated PPD, not a specific currently running WU.

[SandyG]

M for millions… that PPD is out of HFM when those work units were running…

I do not tweak the cards, whatever out of the box they are set for, no voltage changes or the like, I do set an 80 degree limit, but on the 4xxx series never seen anything over 60-65 when I have looked. I have seen 3090 get in the 70s.

I won’t bother giving any more numbers.

All good, my 4090's run at around 62C, the 3090's run into the 80's which is hot, but no A/C in the room with the some of the machines...

Sandy

[Andre_Ti]

Oh, and as the first line of the post said, those were estimated PPD, not a specific currently running WU.

I gave an example of an average PPD of project 18725, and not a specific working unit, but for some reason I was pointed out to receive a small amount of points, which is not true. For monitoring, I use HFM with averaging the results for all control points. Perhaps you were hurt by the value of the PPD different from yours)
For example - https://drive.google.com/file/d/1A5zd-c ... sp=sharing

I won’t bother giving any more numbers.

And rightly so, since you immediately critically perceive the request to indicate your GPU work characteristics (frequency, power consumption). Meanwhile, citing their low temperatures as an example.

All good, my 4090's run at around 62C.

Yes, the temperature is good, but it was not about her.

It is expected to see your negative reactions at the slightest delay in issuing work units and creating conspiracy theories of researchers)