Still can’t get new work units

[Nathan_P]

The same servers that hand out work units also receive them, however the files are larger when they go back. It is being worked on as even more servers are coming on line in the next few days which will help spread the load. There have been issues with one particular project which the team are currently looking into.

[astrorob]

just want to double check that not being able to get WUs for a 32-core threadripper is normal right now. i have 2 GPUs and they are both busy, but unless i kill the FAH background process and restart, i don't get WUs for days. only sometimes when i restart the backend does the CPU receive a WU. i am a little more hesitant to do this with the 2nd GPU now online as it seems like pausing them doesn't necessarily checkpoint on my nvidia gpu and then it seems to have to start over from scratch. so i don't want to lose that work and hence have not restarted in a while.

thanks!

[Joe_H]

Before all of the extra publicity broke and all the extra help rolled in, they were in the process of getting projects set up. The guys who do these for GPUs were a bit farther ahead of the CPU guys. So there are more CPU projects in the works, including some that will do well on large CPU core counts. Until they show up over the next few days, running two CPU slots splitting up the available CPU threads might get you more work.

As for pausing, if you use the advanced control FAHControl you can pause individual slots. Just right click on the slot and a small pop-up window will give options to pause, start, finish and work on idle.

[astrorob]

ok thanks - that would explain what's happening then. i just wish i had beefier GPUs to offer. i did try to change my CPU slot to 16 cores and it failed twice to get any work, so then i tried 8 cores and that also failed a few times so i just set it back to -1 and decided to wait.

yeah i also have FAHControl open and see the options to pause. i cant tell from the windows process viewer what processes are children of what and even cygwin's PS is not showing me that. currently i see two FAHCore_22 processes and two FAHCoreWrappers which i guess correspond to the GPUs? or is one for the GPUs and one for the CPU? there is a single FAHClient process which i guess must be the parent of both of those. i had been pausing everything from the web browser and then killing FAHClient, but maybe that isn't such a good idea.

[Joe_H]

One wrapper for each core process, so when you get a CPU WU there will be another wrapper process and a FAHCore_A7 process running.

Basically it goes like this, FAHClient sits in the background taking care of downloading, uploading, start or stop folding a WU, logging and a few other things. FAHControl and Web Control talk to it over a local to your PC network connection to pass to FAHClient setup and control commands, and monitor information passed from FAHClient.

FAHClient when it has a WU to run creates a separate process to run it on the needed CPU or GPU core inside the wrapper process. Again communication is over a local net connection. The wrapper gets info from the active core process and passes it back to the FAHClient process. The wrapper can usually also pass back that the core has crashed if that happens.

When not folding, FAHClient is designed so it should take very few resources. It is just waiting for some input. On Linux and OS X it is by default set to always be running, limitations on Windows kept it from being setup that way for that OS.

[astrorob]

thanks. that all makes sense. i have basically 0 experience with windows but had to configure this 32-core machine with W10 so it's all new to me. when FAHClient pauses a FAHCore_* process, does that core process eventually exit or does it just sit there waiting for a command to start back up?

[Joe_H]

Yes, the core process should eventually exit. The A7 core tries to write out a checkpoint when it gets a signal to exit, so it can be a short time before it finishes. Not as familiar with the GPU cores (21 & 22) exiting. There are messages in the log that record the final exit message. Something like this from on of my logs:

Code: Select all

17:34:31:WU01:FS00:0xa7:Caught signal SIGINT(2) on PID 11259
17:34:31:WU01:FS00:0xa7:Exiting, please wait. . .
17:34:43:WU01:FS00:0xa7:Folding@home Core Shutdown: INTERRUPTED
17:34:43:WU01:FS00:FahCore returned: INTERRUPTED (102 = 0x66)

As you can see it took a total of ~12 seconds. These are the messages running on OS X, similar ones will show in a Linux log, have not seen the ones in a Windows client log recently enough to remember how they appear.

[mps16au]

is there a way to configure this software.?

[TxRedneck]

By chance is there still a CPU WU shortage? I folded for many years, but stopped for a plethora of reasons. With this whole covid19 focus, I've begun to fold again but am not getting anything for my cpu. Are the WU's specific to different cpu platforms now? I'm unsure of if that's changed in the years between when I used to fold and now. I only ask about the platform thing because a buddy running a lil i3 was able to get a WU, but I'm not on my 1700x.

TIA,
Tx

[Joe_H]

Not so much a shortage, as that servers are at or close to the maximum rate of assigning WUs and receiving them back. More servers are in the process of being put online, setup and projects sent out from them. Basically the number of people/machines have grown about 20x in a period of about 2 weeks.

There are CPU projects that just went from beta test to being released for everyone. You can see the number of WUs available on the servers - https://apps.foldingathome.org/serverstats. As each WU is returned, another is created to be the next generation of the PRCG trajectory.

[TxRedneck]

Not so much a shortage, as that servers are at or close to the maximum rate of assigning WUs and receiving them back. More servers are in the process of being put online, setup and projects sent out from them. Basically the number of people/machines have grown about 20x in a period of about 2 weeks.

There are CPU projects that just went from beta test to being released for everyone. You can see the number of WUs available on the servers - https://apps.foldingathome.org/serverstats. As each WU is returned, another is created to be the next generation of the PRCG trajectory.

[hennings]

Hi, i instantly got new work after restarting FAH, and also setting cpu slots to 8. Whichever it was, i am certainly aware that most of you know this and if does not change the fact, that servers are not catching up with increasing numbers of donors, but if there is a slight chance that the default number of slots prohibited distributing work to my machine, i think that this might still be useful. Mostly for people having their machines running not 24/7 and in the meantime in worst case getting disappointed by their chance of contributing. Not in most cases, but still.

[SparkiTom]

I've had FAH running for a few days now, getting used to it learning how it works. Total noob still obviously.
I've got my CPU and GPU enabled for doing WU's.
In the advanced window, when either CPU/GPU is listed as Ready, if the reset timer is more than 2 or 3 minutes I just right click on it to pause/unpause it to reset that time to less than 90 seconds, or else that reset timer could end up being hours long waiting for the server to give a WU.
No need to reset the whole program especially if either the CPU or GPU is currently working on a WU.