FAH has COVID-19 projects

Gilthanis

[H]ard|DCer of the Year - 2014
Joined
Jan 29, 2006
Messages
8,198
Yeah...my PPD is minimal since they got all of this publicity. I'm keeping those resources busy through BOINC though. I have the BOINC client set up to pause GPU work once I have FAH work units to process.
 

shaggy77

Gawd
Joined
Jul 2, 2005
Messages
763
this just keeps going nuts. Current flops of the F@H network is 1.5 exaflops, a few days ago it was 470 petaflops

https://stats.foldingathome.org/os

My boss tried F@H for the first over the weekend. He's like I now understand what the hell you were talking about at the Friday morning meeting about computing power and Covid19. I think he has 3 rigs going full time. 2 in the office and 1 at home.
 

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
Joined
Jun 27, 2001
Messages
15,696
I'm baaaaaccccckkkk

Screenshot_2020-03-26_15-20-00.png



gonna get a couple more boxes up, lets see how it goes
 

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
Joined
Jun 27, 2001
Messages
15,696
I just installed the client on a box with 40c/80t can we not assign more than 32 cpus to a project?
 

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
Joined
Jun 27, 2001
Messages
15,696
well it has assigned 32 (config errors if you put more than 32) and finally got a WU, and is using 40% of the CPUs as expected... so I may just put another 32 core CPU instance and a 16 CPU instance to use all 80 threads
Screenshot_2020-03-26_19-57-18.png


edit:

seems it doesn't utilize threads only cores?

Screenshot_2020-03-26_20-19-26.png
 
Last edited:

EXT64

DCOTM x3
Joined
Mar 27, 2013
Messages
593
When you aren't using all threads, windows will assign 1 per core before doubling up to maximize performance.

There are units that can use more than 24t, though there are many that cannot. I am currently running one at 128t. The servers are in pretty rough shape right now due to load, so it probably isn't assigning the best units to your system at the moment.

Edit: Since your chips are 10c/20t I would do in multiples of that. I would probably start with 4 20t or 2 20t to start with.
 

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
Joined
Jun 27, 2001
Messages
15,696
so what would be the best configuration in this case?
 

EXT64

DCOTM x3
Joined
Mar 27, 2013
Messages
593
I snuck in an edit - I would recommend multiples of your socket core count.
 

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
Joined
Jun 27, 2001
Messages
15,696
how would I do that though if the client won't let me assign more than 32 CPUs?
 

shaggy77

Gawd
Joined
Jul 2, 2005
Messages
763
Running my new build so far. Finally running on the Prime X570-Pro board with a R7 3700X CPU. This thing just chews on this work. I did run PBO for a day but I was not to happy with the high temps and not much of a gain in speed. I think I ran about 10° C hotter for about 250mhz in clock speed. However the reds in the power usage bar went away when using the PBO. Anyway, Folding the [H]orde.


X570R73700_web.jpg
 

EXT64

DCOTM x3
Joined
Mar 27, 2013
Messages
593
Wow, I wonder if that is a Windows bug. I've never seen that before.

Edit: I'll ask.
 

Nathan_P

[H]ard DCOTM x2
Joined
Mar 2, 2010
Messages
3,431
Ref my EVGA post, i think that may have been due to either WU size or the a4 core that was in use at the time.

As for FLECOM's error Sorry, i can't help with windows, my 2p is on linux. I haven't pushed it beyond 20 threads due to the lack of WU, since we are going into lockdown on sunday I guess I now have something to investigate on my day off!
 

The_Heretic

Certified [H]
Joined
Jun 22, 2001
Messages
13,358
I'm baaaaaccccckkkk

gonna get a couple more boxes up, lets see how it goes

Glad to see you bring some firepower back online. I'm only able to bring a couple of instances back now, myself. But it's still something in the overall scheme.


kurtz4.jpg
 

Icecold

n00b
Joined
Jul 21, 2013
Messages
45
lucky you, my current output is around a third of what it should be
It must have been all my bad luck leading up to it - I was struggling to even get a single WU it seemed like for awhile, but it's gotten better the last 2 days. I just checked and I have a machine sitting idle waiting on a WU again, though.
 

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
Joined
Jun 27, 2001
Messages
15,696
As for FLECOM's error Sorry, i can't help with windows, my 2p is on linux.

no problem, I can move it over to linux next week, box was a spare so it's not doing anything else anyway
 

plext0r

[H]ard DCOTM x3
Joined
Dec 1, 2009
Messages
780
Anyone seeing errors like the following?
17:51:07:WU00:FS00:0xa7:******************************** Build - libFAH ********************************
17:51:07:WU00:FS00:0xa7: Version: 0.0.18
17:51:07:WU00:FS00:0xa7: Author: Joseph Coffland <joseph@cauldrondevelopment.com>
17:51:07:WU00:FS00:0xa7: Copyright: 2019 foldingathome.org
17:51:07:WU00:FS00:0xa7: Homepage: https://foldingathome.org/
17:51:07:WU00:FS00:0xa7: Date: Nov 5 2019
17:51:07:WU00:FS00:0xa7: Time: 06:13:26
17:51:07:WU00:FS00:0xa7: Revision: 490c9aa2957b725af319379424d5c5cb36efb656
17:51:07:WU00:FS00:0xa7: Branch: master
17:51:07:WU00:FS00:0xa7: Compiler: GNU 8.3.0
17:51:07:WU00:FS00:0xa7: Options: -std=c++11 -O3 -funroll-loops -fno-pie
17:51:07:WU00:FS00:0xa7: Platform: linux2 4.19.0-5-amd64
17:51:07:WU00:FS00:0xa7: Bits: 64
17:51:07:WU00:FS00:0xa7: Mode: Release
17:51:07:WU00:FS00:0xa7:************************************ Build *************************************
17:51:07:WU00:FS00:0xa7: SIMD: avx_256
17:51:07:WU00:FS00:0xa7:********************************************************************************
17:51:07:WU00:FS00:0xa7:project: 16402 (Run 0, Clone 248, Gen 10)
17:51:07:WU00:FS00:0xa7:Unit: 0x0000000c96880e6e5e7ebe10f2cdced7
17:51:07:WU00:FS00:0xa7:Reading tar file core.xml
17:51:07:WU00:FS00:0xa7:Reading tar file frame10.tpr
17:51:07:WU00:FS00:0xa7:Digital signatures verified
17:51:07:WU00:FS00:0xa7:Calling: mdrun -s frame10.tpr -o frame10.trr -x frame10.xtc -cpt 15 -nt 24
17:51:07:WU00:FS00:0xa7:Steps: first=5000000 total=500000
17:51:07:WU00:FS00:0xa7:ERROR:
17:51:07:WU00:FS00:0xa7:ERROR:-------------------------------------------------------
17:51:07:WU00:FS00:0xa7:ERROR:program GROMACS, VERSION 5.0.4-20191026-456f0d636-unknown
17:51:07:WU00:FS00:0xa7:ERROR:Source code file: /host/debian-stable-64bit-core-a7-avx-release/gromacs-core/build/gromacs/src/gromacs/mdlib/domdec.c, line: 6902
17:51:07:WU00:FS00:0xa7:ERROR:
17:51:07:WU00:FS00:0xa7:ERROR:Fatal error:
17:51:07:WU00:FS00:0xa7:ERROR:There is no domain decomposition for 20 ranks that is compatible with the given box and a minimum cell size of 1.45733 nm
17:51:07:WU00:FS00:0xa7:ERROR:Change the number of ranks or mdrun option -rcon or -dds or your LINCS settings
17:51:07:WU00:FS00:0xa7:ERROR:Look in the log file for details on the domain decomposition
17:51:07:WU00:FS00:0xa7:ERROR:For more information and tips for troubleshooting, please check the GROMACS
17:51:07:WU00:FS00:0xa7:ERROR:website at http://www.gromacs.org/Documentation/Errors
17:51:07:WU00:FS00:0xa7:ERROR:-------------------------------------------------------
17:51:11:WU00:FS00:0xa7:WARNING:Unexpected exit() call
17:51:11:WU00:FS00:0xa7:WARNING:Unexpected exit from science code
17:51:11:WU00:FS00:0xa7:Saving result file ../logfile_01.txt
17:51:11:WU00:FS00:0xa7:Saving result file md.log
17:51:11:WU00:FS00:0xa7:Saving result file science.log
18:01:07:WARNING:WU00:FS00:FahCore returned: UNKNOWN_ENUM (127 = 0x7f)

I had to move my work directory aside and let the device download new WUs. It was failing like this over and over.
 

EXT64

DCOTM x3
Joined
Mar 27, 2013
Messages
593
Yep, that's the wrapper not being smart enough to catch a thread count that is incompatible with that specific simulation. In the future if you get one of those you can reduce the thread count until it starts folding. I'll report that so they block 24t in the future.
 

Benzino

[H]ard|Gawd
Joined
Mar 3, 2005
Messages
1,581
Getting back into this for the first time in years has prompted some upgrades to deal with cooling and noise:
  • be quiet! DARK BASE 900 case
  • be quiet! DARK ROCK 4 CPU cooler
  • Arctic Accelero Twin Turbo III cooler for the GTX 1070
Doing Rosetta when F@H doesn't have work units available, trying to alternate every day.
 

Benzino

[H]ard|Gawd
Joined
Mar 3, 2005
Messages
1,581
Update: Installed everything and wow, this case is quiet! I hear a low hum at most. CPU temp dropped also, average was 75C now low 60C range on folding load.
Unfortunately I must have messed up the GPU install since F@H causes a hard shut down doing GPU folding. Running Unigine benchmark puts my GPU at 95C but doesn't cause a shut down. So, I have some new Grizzly thermal paste and some aluminum heatsinks w/thermal tape that arrived today. Going to spend some more time on Youtube videos to see if I can get the GPU cooling straightened out before going back to F@H.

In the meantime I've been cranking out Rosetta and don't even notice. The be quiet! gear was $$$ but worth it.
 

Xilikon

[H]ard|DCer of the Year 2008
Joined
Oct 12, 2004
Messages
14,523
Hello everyone from another oldtimer!

I fired up the new client on my Dell XPS 9570 laptop to contribute for the COVID-19 fight. I'm pleased to see a few old timers coming back as well like FLECOM :D
 

Xilikon

[H]ard|DCer of the Year 2008
Joined
Oct 12, 2004
Messages
14,523
Isn't it normal that the v7 FAH web control show 1,730,491 points earned when in the FAH stats for mine, it's 34,142,115 ? If I check for team stats, it display correctly and I wondered where is the issue
 

Benzino

[H]ard|Gawd
Joined
Mar 3, 2005
Messages
1,581
I initially didn't install the GPU cooler correctly and had hard shutdowns when GPU folding (hitting 90C) had to stop for a while. Finally got some Grizzly thermal paste and some aluminum heatsinks and reassembled, fodling quite nicely now. At 58C under GPU load now and stable. The orientation of the GPU in my case makes me wonder if the aluminum heatsinks are going to fall off, the thermal strips don't strike a lot of confidence in me. I might have to look in the case again to see what might have fallen off, may need to skip the thermal tape and go with thermal glue.
 
Top