LTT Folding Team's Emergency Response to Covid-19

[TrueMad]

Got my 1st CPU WU from linus1.foldingathome.org

 

Quote

13:37:38:WU01:FS00:Connecting to 65.254.110.245:8080
13:37:38:WU01:FS00:Assigned to work server 168.245.198.125
13:37:38:WU01:FS00:Requesting new work unit for slot 00: READY cpu:10 from 168.245.198.125
13:37:38:WU01:FS00:Connecting to 168.245.198.125:8080
13:37:39:WU01:FS00:Downloading 1.23MiB
13:37:40:WU01:FS00:Download complete

 

[HaRdLy007]

4 minutes ago, SCSI15 said:

Fans or gpu usage?

Both, the core drops from 2060MHz to around 1400MHz and the fans slow down.

[axpere]

When will LTT do a follow up video on the work server installation?

[LFreeman]

4 minutes ago, HaRdLy007 said:

Both, the core drops from 2060MHz to around 1400MHz and the fans slow down.

sounds like the fan curve is poo poo and is thermal throttling. try setting fans to 90% and keep an eye on if the clocks drop.

[bafo_ah]

4 minutes ago, HaRdLy007 said:

Both, the core drops from 2060MHz to around 1400MHz and the fans slow down.

When GPU drop, are CPU rise? Maybe CPU bottleneck? GPU fold still need CPU to supply "block of data" (I think)

[LFreeman]

1 minute ago, bafo_ah said:

When GPU drop, are CPU rise? Maybe CPU bottleneck? GPU fold still need CPU to supply "block of data" (I think)

Not a bad assumption either..

[ShortRouter]

25 minutes ago, Mike The Fool said:

Yeah, I do this manually when I'm up and around, just haven't yet mastered the ability to interact with my computer from the dream realm

I use TeamViewer free version with unattended access set-up + mobile TeamViewer App to check on my PC. Works great both locally and over the internet.

If only access over the local network is needed: https://community.teamviewer.com/t5/Knowledge-Base/Can-TeamViewer-be-used-within-a-local-network-LAN-only/ta-p/4618

[Inkertus]

6 minutes ago, LFreeman said:

Not a bad assumption either..

F@H sits somewhere between crypto crunching and video games in terms of how much it needs to actively maintain and send data to the GPU to keep it maximally busy.  I've learned it needs one thread per GPU to do so.  Not sure if it needs more than one if your CPU is lower-end.

[bafo_ah]

5 minutes ago, LFreeman said:

Not a bad assumption either..

Just look at mine, when my GPU drop (GPU 1), CPU rise

Spoiler

GPU1: Hey CPU what take you so long?

CPU: Nah, I just give you time to take rest for folding all day

 

[ShortRouter]

30 minutes ago, HaRdLy007 said:

Anyone else using a 5700XT? Not sure if this is normal, but it will ramp up for about 30 seconds and then will ramp down for about 60 seconds.

Delete CPU slot in FAHcontrol and observe if this still happens. If it does it's most likely a fan curve / overheat issue. if it solved the problem, you have to set some parameters for CPU folding slot to not use all cores available. Or just stick to folding on GPU only.

[shaz2sxy]

Totally forgot about this event! Ive been working from home, so all over the place.

 

Wont be able to get all my systems up as son at home so using his PC to play

 

will just have my 2070 super when im not working

 

but every little helps

[bafo_ah]

1 minute ago, shaz2sxy said:

will just have my 2070 super when im not working

You wont use 2070 super for "work"

[ShortRouter]

3 minutes ago, bafo_ah said:

You wont use 2070 super for "work"

Why not? I edit videos daily on an RTX 2080 (technically a 2070 Super with little more CUDA cores).

[Donut Dan]

7 hours ago, Justaphf said:

@Donut Dan Ran into this when moving cards between systems.  The system I removed a card from I had to manually edit the config file and delete the extra GPU slot.

 

On Windows the config file was located at:

   C:\Users\Justaphf\AppData\Roaming\FAHClient\config.xml

Type %APPDATA%\FAHClient\ in Windows Explorer address bar to easily reach this location as the path is based on the user you logged into Windows with.

 

EDIT: If you are on Linux, the default location for the config file is:

   /etc/fahclient/config.xml

 

So if I removed one of my two cards, the config was below, and I would delete the slot with id='2'.  Not sure which slot matters, on a different one I deleted the first slot and there was a gap and it still worked.  Save the changes to the config file and restart the FAHClient.  Hope that helps!

Thanks a million, some toying and retyping and it's good to go!

[Donut Dan]

once again, the folding site is certain that I don't exist

 

[cbigfoot]

well iim up and happy all my computers are chugging except for my R7-350x which i wont cry over...

[shaz2sxy]

8 minutes ago, bafo_ah said:

You wont use 2070 super for "work"

haha what i mean by "work" is gaming... just don't tell my boss

[cbigfoot]

 

9 hours ago, Mithras Pan said:

I'm running the same case and it's a shame to hear that they stopped making it.

Yeah my only complaint is the Front has to come off for the filter...i am also about to take it down to swap the fans to a Push config to see if it works any better.

8 hours ago, Gorgon said:

LMFAO - so i added a GPU slot for the 5670 and it assigned me to a WS IP 192.0.2.1 which is a reserved, non-routable IP address

yeah i think that is their way of saying we dont need that antique

[Yujah]

1 hour ago, GOTSpectrum said:

I know the HD 7950 is supported from personal experience, but can't comment on anything older than that

Yes. the discriminator is double-precision floating-point math support. The 7950 is GCN and all GCN chips should work. The HD 5000 and HD 6000 series, some HD 7000, are still TeraScale though and need an HD 5830+ (TeraScale 2) or HD 6930+ (TeraScale 3) to be usable by current F@h.

 

HD 7730+ is GCN1 and definitely works, before that HD7000 branded cards are TeraScale 3 (discrete cards) or TeraScale 2 (APUs) but that should also work it seems. I.e., 5830+, 6930+.

 

Specifically then, an HD 5750 will indeed not work. Note that this will/would not be evident in the client itself but that you'd getting 192.0.2.1 assigned as a work server each time. That's an unroutable, reserved IP address and F@h's way of saying "hell no!" at runtime...

[TheDude92]

38 minutes ago, axpere said:

When will LTT do a follow up video on the work server installation?

According to their WAN show stream on Friday, they should be launching it this weekend and having an update on it by Monday.

[shaz2sxy]

just sitting here waiting for a WU.... COME ON lol

[cbigfoot]

11 minutes ago, Donut Dan said:

once again, the folding site is certain that I don't exist

 

You exist it took a while to find you but i did...

https://apps.foldingathome.org/teamstats/team223518.html

[Favebook]

1 hour ago, Saimen said:

I don't know, how do I find out?

If you do not know what beta flag is or if you are running it, DW, you are not running it. It's an additional option in Expert options which Beta testers use.

 

They test WUs which pass Alpha phase, and then if it passes Beta standards, they pass it to Advanced after which it goes to public (no flag needed - this is what most users run).

[Shigeru]

11 minutes ago, Yujah said:

Yes. the discriminator is double-precision floating-point math support. The 7950 is GCN and all GCN chips should work. The HD 5000 and HD 6000 series, some HD 7000, are still TeraScale though and need an HD 5830+ (TeraScale 2), HD 6930+ (TeraScale 3), HD 7730+ (GCN1) to be usable by current F@h.

 

Specifically then, an HD 5750 will indeed not work. Note that this will/would not be evident in the client itself but that you'd getting 192.0.2.1 assigned as a work server each time. That's an unroutable, reserved IP address and F@h's way of saying "hell no!" at runtime...

Indeed, that was clear for my 8400 GS after the first 192.0.2.1, but thanks for the additional information!

[axpere]

10 minutes ago, TheDude92 said:

According to their WAN show stream on Friday, they should be launching it this weekend and having an update on it by Monday.

Thanks!