LTT Folding Team's Emergency Response to Covid-19

[Alex Atkin UK]

Quote

23:18:01:WU01:FS01:0x22:Folding@home Core Shutdown: FINISHED_UNIT
23:18:02:WU01:FS01:FahCore returned: FINISHED_UNIT (100 = 0x64)

...

01:57:12:WU01:FS01:Sending unit results: id:01 state:SEND error:NO_ERROR project:13878 run:0 clone:1808 gen:28 core:0x22 unit:0x0000002434e06d4a5e80cfe2758eb913
01:57:12:WU01:FS01:Uploading 48.06MiB to 52.224.109.74
01:57:12:WU01:FS01:Connecting to 52.224.109.74:8080
01:59:23:WARNING:WU01:FS01:WorkServer connection failed on port 8080 trying 80
01:59:23:WU01:FS01:Connecting to 52.224.109.74:80
02:01:34:WARNING:WU01:FS01:Exception: Failed to send results to work server: Failed to connect to 52.224.109.74:80: Connection timed out

I have two WUs sat waiting trying to send to this collection server now.

[Grounded]

1 hour ago, LAR_Systems said:

That depends on when you ask

So currently the PPD on the APU is sucking lemons due to the work unit... so total PPD is 400k... which is terrible

The PPD of the 1060 is consistently around 320k on it's own...  so never bank on the APU it's luck of the draw, discrete GPUs are still the way to go.

Sweet thanks for the info. Hopefully I can get my setup somewhere close to around that

[cbigfoot]

39 minutes ago, Grounded said:

I messaged them yesterday but still haven't heard back yet.

give him a day or 2 hes only 1 man and has cat issues...must play with cat or be destroyed...

[Bitter]

4 hours ago, Bitter said:

I got my Linux/P106 issue sorted, it was a config issue with FAHControl. Now I just can't get GPU work units to come through at all for any of my PC's, did they run out again?

And this I broke everything again messing with trying to get coolbits enabled to give me fan control so I reinstalled Ubuntu rather than trying to fix what I broke because it's faster to just start over. This time screw the coolbits and fan control.

 

I am however having an odd issue with opencl, I'll post the logfile and tag some people tomorrow. It finally found a work unit to run so once it's working and working well I'll get back at the opencl issues. It seems like it just didn't install opencl properly when the installer was setting the system up, so I'll purge nvidia then reinstall from terminal and see how that goes before anything more nuanced. linux, just hit it with a hammer till it works.

[Erasmuz]

Sorry for asking what I think is a silly question - but am I right in thinking this runs better under Linux? Is there anything more to it compared to the Window's client? Do I just download the .DEB & install?

 

I dual-boot Windows & Ubuntu, I happened to be in Windows when I decided to join so installed the F@H client there. It took less then a minute and it's up & running, but if it runs significantly better under Linux I'll switch over - although I'm not keen enough to spend a heap of time optimizing things.

[efka112]

54 minutes ago, Erasmuz said:

Sorry for asking what I think is a silly question - but am I right in thinking this runs better under Linux? Is there anything more to it compared to the Window's client? Do I just download the .DEB & install?

 

I dual-boot Windows & Ubuntu, I happened to be in Windows when I decided to join so installed the F@H client there. It took less then a minute and it's up & running, but if it runs significantly better under Linux I'll switch over - although I'm not keen enough to spend a heap of time optimizing things.

for me  its similar on both operating systems, so think you can leave it as it is

[NetoriusNick]

I have not had any issues running it headless. Only "complicated" part was the config file to allow remote access/control, this is because I manually edited the config file.

[Alex Atkin UK]

2 hours ago, Bitter said:

And this I broke everything again messing with trying to get coolbits enabled to give me fan control so I reinstalled Ubuntu rather than trying to fix what I broke because it's faster to just start over. This time screw the coolbits and fan control.

As I recall all I did (on Fedora though) was use nvidia- settings to save the current configuration as ~/xorg.conf, copy that into /etc/X11/xorg.conf.d/10-nvidia.conf then add one line:

Quote

Section "OutputClass"
    Identifier "nvidia"
    MatchDriver "nvidia-drm"
    Driver "nvidia"
    Option "AllowEmptyInitialConfiguration"
    Option "PrimaryGPU" "yes"
    Option "SLI" "Auto"
    Option "BaseMosaic" "on"
    Option "Coolbits" "28"
EndSection

 

Its rather more important for my card though as for some bizarre reason the stock fan profile is useless, its causes severe thermal throttling as it only uses about 40% fan at 83C despite the fact you only need 50% to maintain stock clocks.

[Favebook]

Here's one longer video where they explain a lot:

 

 P.S.

Starts at around 00:05:00 if I remember correctly, so be free to skip first few minutes.

[Alex Atkin UK]

Finally submitted, its frustrating to only get half the points though due to server issues.  Ah well, the important thing is the work was submitted before the timeout so wasn't a waste.

[Bombastinator]

For what it’s worth differences apparently have been made.  More needs to be done though.  The guardian just did an article on F@h.  Apparently those WUs were modeling how the corona spike opens to attach to a cell.  They got stage 1 with f@h.  There are apparently several stages

 

[Kilrah]

Yeah an animation of that was posted 2 weeks ago already. 

[Bitter]

10 hours ago, Alex Atkin UK said:

As I recall all I did (on Fedora though) was use nvidia- settings to save the current configuration as ~/xorg.conf, copy that into /etc/X11/xorg.conf.d/10-nvidia.conf then add one line:

Its rather more important for my card though as for some bizarre reason the stock fan profile is useless, its causes severe thermal throttling as it only uses about 40% fan at 83C despite the fact you only need 50% to maintain stock clocks.

I may have another go at it tonight, if it's not folding. I actually wanted to turn the fans down at idle, don't need 40% to idle at exactly ambient temp lol, it can idle a little warmer with slower fans. I can at least change the power limit so I don't power throttle, but I haven't been able to get enough worthwhile work to see if that's actually a problem yet. The WU I got last night was worth like 20,000 and was going to take 6 hours.

[ShortRouter]

6 hours ago, Bombastinator said:

For what it’s worth differences apparently have been made.  More needs to be done though.  The guardian just did an article on F@h.  Apparently those WUs were modeling how the corona spike opens to attach to a cell.  They got stage 1 with f@h.  There are apparently several stages

 

Protein research is just laying the foundation for developing a vaccine. Many testing and certification phases is what takes ages but needs to be done. Animal testing, then human testing on small groups, looking for possible side effects, etc etc. Friend of mine is a senior research manager at Ratiopharm, that's how I know. To put it mildly, scientists don't know shit yet about how proteins work exactly so they shoot at the problem with bruteforce raw compute power simulating millions of possible outcomes and then concentrate on those that could be useful to "inject" something into the "spike" protein for example.

[Cabanaj]

Hmm, I think the Folding Gods are sending us a message. In the last 24 hours, I have received non stop CPU Working Units, and for every 3 CPU WU, I got one GPU WU.
I'll have to switch that stock cooler. I appreciate the heat source in the cold nights, but the plane-like noise... Noctua NH-D15 it is! 
 

[fangfufu]

I WON A STEAM KEY!!!

[GrockleTD]

4 minutes ago, fangfufu said:

I WON A STEAM KEY!!!

congrats!

[Bitter]

Who else can't seem to get their completed units to upload?

[efka112]

7 hours ago, Bitter said:

Who else can't seem to get their completed units to upload?

you get message in log that upload failed or just don't get points?

[Kilrah]

Had one stuck for half an hour yesterday, transfer started and failed at a random percentage, looking at server stats it was rebooting constantly, uptime kept varying between 1-2-3 minutes and back... eventually stayed up and took my WU.

[marknd59]

 

7 hours ago, Bitter said:

Who else can't seem to get their completed units to upload?

 

I have 3 that have been stuck for over a day.

[efka112]

1 hour ago, marknd59 said:

 

 

I have 3 that have been stuck for over a day.

if WUs stuck for a while, check server status if server is online and have disk space, try to relaunch FAH. sometimes is issue from client side  

[FraktalU]

1 hour ago, marknd59 said:

 

 

I have 3 that have been stuck for over a day.

Also got one that is stucked since yesterday 14h

[marknd59]

13 minutes ago, efka112 said:

if WUs stuck for a while, check server status if server is online and have disk space, try to relaunch FAH. sometimes is issue from client side  

 I've got 2 stuck trying to send to 155.247.166.219 which is up and down like a yoyo and only has 2Tib left on it so it's pretty full up, and 1 stuck that lists the work server as 13.82.98.119 and collection as 0.0.0.0. I've tried shuting down and restarting F@H and all that did was get the retry time down, not a lot I can do about it.

[Krankenstein]

GPU not getting any more folding work so I am cutting my folding for now. GPU sits at around 300 attempts and got nothing. (Yes I did the pause/fold clicking). Next attempt is in 14hrs.

CPU gets hit by these small work units that seem very inefficient.

I'll be back when the next event for F@H is scheduled!