Anandtech vs Tom's Hardware Folding@Home Coronavirus Race thread

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

StefanR5R

Diamond Member
Dec 10, 2016
3,444
3,696
106
Which work server and collection server is this WU supposed to go to?

Try a reboot. This often helps to get stuck uploads going again.
 
  • Like
Reactions: JC

JC

Diamond Member
Feb 1, 2000
5,744
7
81
Just saw that advice from earlier in the thread. I should pay more attention :eek:

Thanks!
 

Assimilator1

Elite Member
Nov 4, 1999
23,444
193
106
Fingers crossed, what's azure anyway? ;)

Btw, I shall be switching CPU power to Rosetta when SETI hibernates next week.

Btw re THG & BOINC, they've got a SETI team there though haven't they??
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,276
7,922
136
Fingers crossed, what's azure anyway? ;)

Btw, I shall be switching CPU power to Rosetta when SETI hibernates next week.

Btw re THG & BOINC, they've got a SETI team there though haven't they??
GPU should go to F@H. CPU should go to Rosetta. They are mutually exclusive (in what they do or do best)
 

NesuD

Diamond Member
Oct 9, 1999
4,997
105
106
One WU in the wee hrs of the morning and 2 more completed earlier this evening that for some reason didn't show up in the last stats update. Just got another machine online and it started working almost immediately although at a much slower rate than my main which is just sitting idle most of the time.
 

StefanR5R

Diamond Member
Dec 10, 2016
3,444
3,696
106
Relative drop in credits from Sunday to Monday:
TeAm AnandTech: -25.5 %​
Tom's Hardware: -26.5 %​
From Sunday to Tuesday:
TeAm AnandTech: -38 %​
Tom's Hardware: -24 %​

Though we still produced 9 M more than they did yesterday.

Hmm, I woke up to 4 WUs downloading, 6 WUs running, and 8 WUs sending. Time to reboot.

--------
Edit, the EOC stats site has news:
As promised, the aggregate stats are up! ...
you can find a link on the left sidebar where it says, All Summary, or on the All Teams List, I put the "Aggregate Teams" up top with rank 0.
And this is how the daily production of all F@h contributors combined looks like:

March_24_combined.png

Also from the combined page:
Active users: 145,466 (+85,296)​
New users: 99,310 (+47,219)​
Avg PPD: 72,831.8​

TeAm AnandTech on the All Teams List:
17th by points ever produced,​
14th by number of active users,​
7th by points during this week! (since Sunday, March 22 inclusive)​
 
Last edited:

Shmee

Memory and Storage, Graphics Cards
Super Moderator
Sep 13, 2008
4,314
446
126
Does F@H work well on AMD cards now? I could throw the Vega 64 and 5930k rig at it.
 

VirtualLarry

Lifer
Aug 25, 2001
48,155
5,007
126
Does F@H work well on AMD cards now? I could throw the Vega 64 and 5930k rig at it.
Yes, I believe so, although CPU folding has a sort of diminishing-returns aspect to it, compared to GPU folding. Like Markfw said, put the CPU on Rosetta@Home (BOINC), and the GPU on F@H.
 

StefanR5R

Diamond Member
Dec 10, 2016
3,444
3,696
106
It may be a good time to switch at least one of them over to Linux. My main hurdle before was I couldn't get FAHControl to work because of it needing python-gnome2 which isn't part of recent Linux distro's. Is everybody just running FAHControl on another(Windows?) machine to connect to all their Linux machines, or there's an easy workaround I wasn't finding regarding python-gnome2?
(Emphasis mine. For reference, @biodoc's answer, referring to Linux Mint:)
FAHControl works on linux mint 19.2 and 19.3.

After installation of linux mint:
sudo apt update
sudo apt upgrade
reboot
go to menu , administration, device manager to install nvidia driver then reboot
sudo apt install ocl-icd-libopencl1
sudo apt install ocl-icd-opencl-dev
Edit: Install Fahclient and FAHConrol
cd /var/lib/fahclient
sudo wget http://fah-web.stanford.edu/file-releases/public/GPUs.txt

Now restart the FAHClient:
sudo service FAHClient restart
(And @Soulkeeper's, re Slackware:)
I use slackware.
I had to create my own package for fah based on the ubuntu/debian ones.
Just like I had to create my own ROCm packages from source.

If any fellow slackware users need help let me know.
The foldingforum has got a procedure for several *.deb based distributions:
FAHControl on Ubuntu 19 / Debian 10 and variants.
Though two users reported breakage of standard packages from this procedure, requiring them to revert these changes and leaving the FAHControl issue on Ubuntu or Debian unresolved for them.

--------

Does F@H work well on AMD cards now? I could throw the Vega 64 and 5930k rig at it.
Navi had issues for a while, but they should be mostly resolved by now. (In the thread, FahCore_22 is mentioned as "beta" work which needs to be enabled explicitly in the client. It was promoted out of beta status meanwhile, so this config should no longer be necessary.)
Vega, Polaris, GCN etc. should just work.

Folding on AMD GPUs on Windows should be straightforward. On Linux, you may have to install OpenCL enabled drivers as a 3rd party package first.
 

mopardude87

Platinum Member
Oct 22, 2018
2,793
1,098
96
Ugh, was hoping more people wouldn't have issues. I always run into issues with folding, usually quit cause its more trouble then its worth. I would think in 2020 that folding@home would just work by now.
 

Assimilator1

Elite Member
Nov 4, 1999
23,444
193
106
Don't forget that theirs been a massive influx of crunching power to F@H which has overwhelmed their servers (more have been & are being added).
It isn't normally this bad AFAIK, although I'm only a part time folder, I'm sure some of the full time folders can confirm or deny this :).

GPU should go to F@H. CPU should go to Rosetta. They are mutually exclusive (in what they do or do best)
As per my earlier posts I'm already folding on my GPU ;).... when I can get WUs!
 

borandi

Member
Feb 27, 2011
133
104
116
Ugh, was hoping more people wouldn't have issues. I always run into issues with folding, usually quit cause its more trouble then its worth. I would think in 2020 that folding@home would just work by now.
So you think going from 40k active users to 150k active users, including a GPU array of 6000 CPUs and a large cloud firm with xx,xxx GPUs, was predictable?
No-one expected all these people to jump in, least of all FAH. They're working with Microsoft's Azure cloud team to build in more capacity.
They already spun up 3 x 100 Terabyte servers so send/recieve work. We've already filled them up in a few days with results.

'I would think in 2020 that FAH would just work'

It does, for 40k people. But then you try supplying everyone when your demand goes up 3x overnight.
 
  • Like
Reactions: Assimilator1

StefanR5R

Diamond Member
Dec 10, 2016
3,444
3,696
106
Ugh, was hoping more people wouldn't have issues. I always run into issues with folding, usually quit cause its more trouble then its worth. I would think in 2020 that folding@home would just work by now.
Well, it does just work.

Two caveats:
  • (ninja'd by @Assimilator1 and @borandi)
    There are so many clients online now, that they are idle about half the time because the bottleneck (previously: contributed clients) has shifted to the workunit servers (mainly: storage bandwidth).
    Though strictly speaking, it works as designed. If work is available, your client gets it. If not, it's waiting until it can get some. Ditto, if the corresponding server is ready to receive the result file, the client sends it. If not, it sends it later once the server becomes ready.
  • If you prefer to fold on GPUs, then the driver, the clock and voltage config of the GPU, the cooling provided, and the power supply need to be up to the task. This applies to any GPGPU application; F@h is no exception here.
    But in practice, a consumer PC can usually be deployed as-is for Folding@home duty, even though it was not designed for work like this. It's a good thing that it's usually possible to take the side panel off. ;-)
 

Icecold

Senior member
Nov 15, 2004
345
25
91
FWIW - using the guidance given in this thread I was able to get Folding@home working on a couple Linux Mint machines. I had been trying Ubuntu before this thread and it was giving all kinds of issues with python-gnome2 and none of the workarounds I could find seemed to help(so I switched back to Windows), but by switching to Linux Mint and following biodoc's instructions exactly as written it is working well. I noticed a PPD bump compared to Windows as well.

I've been having the same issues with getting consistent WU's regardless of the number of times I pause/start, or restart FAHClient, etc. so I've moved my GPU's to GPUGrid until that gets ironed out since it seems like there are more than enough resources being thrown at Folding@home right now.
 
Nov 18, 2016
42
39
61
I meant to ask earlier, I take it folding on a laptop is a bad idea?
yes and no. ultraportables and 2in1 are a no-go from the start. of course any contribution helps, but at this time, like it's been said, we're in no shortage of powerful systems. those laptops would contribute very little and yeah, they could be in danger of overheating if not solidly designed. at the very least, they would throttle heavily, further reducing your folding power. i know that smaller systems get smaller WUs, but at this point, it's actually a greater contribution to not add to the folding pool if the machine can't process a WU in less than, say, 4-6 hours anyway. storage and network bandwidth for stanford is much more valuable at this point than raw compute power.

as for the general rule of thumb, for heavy compute the number and speed of the compute units is king, with memory bandwith second and cpu power last (for GPU based stuff like FaH). there is also a definite spike in PPD per watt for Turing GPUs. (you may want to look up AMD yourself). so yeah, if you got a gaming laptop who's built for sustained, top speed use, say, with a 2060 or 1660Ti/Super, then yes, it can help. older nvidia laptops can also, but with diminishing returns of course. turing is that much more efficient. anything else would be of relatively little use, put the machine in danger (maybe not the chips themselves, but if the solder job is crappy and you manage to desolder something, it's RIP) and so it's best left alone. those laptops are meant for burst speed, not sustained crunching.

that said though, many of them, especially the brand new ryzens and ice lake i5/i7 should pack a decent punch on rosetta, albeit with the same thermal caveats listed above. worth a try maybe, just download some monitoring software like afterburner to closely check the clock and temp behaviour of the chip.
 
  • Like
Reactions: Assimilator1

Assimilator1

Elite Member
Nov 4, 1999
23,444
193
106
I don't have a laptop.......less than 10yrs old anyway! ;)
I was asking as I posted in a family what's app group about folding@home doing research for C19, I stated I doubt that F@H could be done on a laptop, but I would find out, now I have :).
My rigs are in my sig, atm only main rig is crunching (which is usually the case). I might fire up my 2nd rig for the last week of SETI@H being alive for a little boost :) (seeing as my main rig's GPU is folding, when it can)....doh! This is SETI's last week! Firing up 2nd rig now........
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,276
7,922
136
Last 24 hr.

Team Anandtech 97 m
Toms Hardware 100 m

me 6m . not good

But a definite downturn for everybody today., and we are loosing.
 
  • Like
Reactions: Assimilator1

Pokey

Platinum Member
Oct 20, 1999
2,638
206
106
Do we have an official stats master posting the numbers?

I ask because it looks to me like we are ahead in total points and in 24 hr. avg production. How are we losing and by how much?

Just asking.
 

ASK THE COMMUNITY