Interested in a Folding@Home Coronavirus Race?

[TennesseeTony]

For all the ones just joining, or returning to Folding at Home, there are a few things to know about getting the most out of your GPU.

1. Linux Operating System will give you a significant boost in performance, but also in power usage. Windows for some reason only uses 80-85% of the card, IIRC.
2. Make sure the GPU has at least one tread, preferably one cpu core, as some cpu is utilized to feed the GPU.
3. As for points, having a passkey and having completed 10 tasks successfully gets the best points, as well as the Fast Return Bonus, but this does not actually mean more science is getting done

• Other than that, the type of task you receive is generally random, you may be crunching for cancer, then next task Covid-19
• Folding at Home can be a bit finicky, buggy even. Stuck downloads are common. Server issues are common. Please don't let this discourage you though!
• TeAm AnandTech's team number is 198, if you want to crunch for the TeAm. We are 17th by rank, in the world. Followed by Tom's Hardware in 18th.

 

[StefanR5R]

In a sense, returning results swiftly does mean more science getting done. It is not like PrimeGrid for example where a search space can be planned ahead for many months, if not many years. The researchers set up batches of tasks, need to wait for the results of the batches and analyze them further, and based on this decide on new batches. Therefore not just the quantity of work done in a given time frame matters to the researches, but also to keep those time frames short and somewhat deterministic. --- At least that's how I imagine things work at F@h.

Back at a time when my work more frequently involved numerically heavy physics simulations, I generally arranged for a batch of simulation runs with a set of parameter variations to run from the evening until the morning, then during the work hours had my hands full to post-process the results and wrap my mind around what I got to see, then decided on what to simulate in the next go. Tuning the problem sizes and number of problems to fit into this daily cycle on the hardware available to me was part of this work.


(PS, on the other hand, F@h's primary mechanism for quick turnaround is that the client doesn't have a work buffer like boinc does. Maybe the quick return bonus is more an incentive to contributors to put their latest and greatest hardware to use...)

 

[Markfw]

Not much of a race from my side. I still get a variety of errors, most of my fleet is still out of work. Here is a new one:

13:34:48:WU00:FS01:Connecting to 140.163.4.241:80
13:35:11:ERROR:WU00:FS01:Exception: 10002: Received short response, expected 512 bytes, got 0
13:50:33:WU00:FS01:Connecting to 65.254.110.245:8080
13:50:34:WU00:FS01:Assigned to work server 140.163.4.241
13:50:34:WU00:FS01:Requesting new work unit for slot 01: READY gpu:0:GP102 [GeForce GTX 1080 Ti] 11380 from 140.163.4.241
13:50:34:WU00:FS01:Connecting to 140.163.4.241:8080
13:52:13:ERROR:WU00:FS01:Exception: 10002: Received short response, expected 512 bytes, got 0

 

[biodoc]

Not much of a race from my side. I still get a variety of errors, most of my fleet is still out of work.

The lack of work does not bode well for a race.

 

[Markfw]

And the latest error

15:01:36:WU00:FS00:Connecting to 140.163.4.231:8080
15:01:57:WARNING:WU00:FS00:WorkServer connection failed on port 8080 trying 80
15:01:57:WU00:FS00:Connecting to 140.163.4.231:80
15:02:19:ERROR:WU00:FS00:Exception: Failed to connect to 140.163.4.231:80: A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.
16:17:37:WU00:FS00:Connecting to 65.254.110.245:80
16:17:37:WU00:FS00:Assigned to work server 140.163.4.241
16:17:37:WU00:FS00:Requesting new work unit for slot 00: READY gpu:0:GP102 [GeForce GTX 1080 Ti] 11380 from 140.163.4.241
16:17:37:WU00:FS00:Connecting to 140.163.4.241:8080
16:17:58:WARNING:WU00:FS00:WorkServer connection failed on port 8080 trying 80
16:17:58:WU00:FS00:Connecting to 140.163.4.241:80

 

[Ken g6]

Whenever I get a downloading error, I reboot and it tries for another WU. I am currently lucky enough to have both my GPUs working.

 

[Markfw]

Whenever I get a downloading error, I reboot and it tries for another WU. I am currently lucky enough to have both my GPUs working.

After rebooting many times, I am now up to 8 working, 4 idle. Progress. Still have a 2080ti idle, that over 2 million ppd alone !

 

[HutchinsonJC]

Lets see how much this jacks up my home electricity bill, lol, any tips on maximizing efficiency?

If you still have the heat running (Northern Hemisphere) it really won't mess with your electric bill too much unless your primary heat source is something other than electric. And even then, your thermostat has no choice but to recognize the heat from your PCs as part of the whole. So if you're getting 250 watts of heat from a computer, your heat source, regardless of what it comes from, has 250 watts worth of heat less to have to dump in the house.

 

[Markfw]

If you still have the heat running (Northern Hemisphere) it really won't mess with your electric bill too much unless your primary heat source is something other than electric. And even then, your thermostat has no choice but to recognize the heat from your PCs as part of the whole. So if you're getting 250 watts of heat from a computer, your heat source, regardless of what it comes from, has 250 watts worth of heat less to have to dump in the house.

The only problem there, I have gas heat, much more efficient than electric. If he has gas, then the electric bill will go up, but the gas will only go down a little.

 

[Ken g6]

After rebooting many times, I am now up to 8 working, 4 idle. Progress. Still have a 2080ti idle, that over 2 million ppd alone !

Here's a new error, uploading!

19:36:28:WU00:FS02:FahCore returned: FINISHED_UNIT (100 = 0x64)
19:36:28:WU00:FS02:Sending unit results: id:00 state:SEND error:NO_ERROR project:11745 run:0 clone:6823 gen:0 core:0x22 unit:0x000000028ca304f15e6bc3b9601ce3fc
19:36:28:WU00:FS02:Uploading 33.37MiB to 140.163.4.241
19:36:28:WU00:FS02:Connecting to 140.163.4.241:8080
19:38:38:WARNING:WU00:FS02:WorkServer connection failed on port 8080 trying 80
19:38:38:WU00:FS02:Connecting to 140.163.4.241:80
19:38:41:WU00:FS02:Upload 0.19%
19:38:52:WU00:FS02:Upload 0.37%

I'll try a reboot and hope that fixes it.

 

[StefanR5R]

Or wait for the beginning of the workweek, when the admins will be having a chance to implement better support for the current surge in participation than they could during the weekend. :-)

 

[Markfw]

Here's a new error, uploading!

19:36:28:WU00:FS02:FahCore returned: FINISHED_UNIT (100 = 0x64)
19:36:28:WU00:FS02:Sending unit results: id:00 state:SEND error:NO_ERROR project:11745 run:0 clone:6823 gen:0 core:0x22 unit:0x000000028ca304f15e6bc3b9601ce3fc
19:36:28:WU00:FS02:Uploading 33.37MiB to 140.163.4.241
19:36:28:WU00:FS02:Connecting to 140.163.4.241:8080
19:38:38:WARNING:WU00:FS02:WorkServer connection failed on port 8080 trying 80
19:38:38:WU00:FS02:Connecting to 140.163.4.241:80
19:38:41:WU00:FS02:Upload 0.19%
19:38:52:WU00:FS02:Upload 0.37%

I'll try a reboot and hope that fixes it.

I am now down to 6 working, 6 trying to get a unit. I added to a thread here on the folding forum. They are adding servers and capacity. I even volunteered one of my EPYC boxes to assist.

Client failing to connect to assigned work server - Folding Forum

foldingforum.org

This assist was in a PM, so you won't see that

 

[Markfw]

@Ryan Smith , I think we need to let Stanford figure out their server situation before we start the race. They have all sorts of issues at the moment.Only half of my cards have work, and I can't connect with the other half. And it changes every few hours. Maybe tomorrow or Tuesday we can get things up and running. When was the start going to be ????

 

[borandi]

So for what it's worth, it looks like a major cloud computing company is donating time on its GPU farm to the CoVID project. As a result, the work server ran out of work.
They've put a second server up to send out work, but both servers have got 'WU Generation' at a higher priority than 'send out WUs' for now.

They've also been in the process of upgrading one of their 15 TB results servers to 2x 100 TB results servers. It shouldn't be an issue for now, but could be. I'm uploading results of the WUs I managed to get very slowly right now. Of course, there's no work for the GPU to do after.
I suspect on Monday they'll have a couple of people in the lab focusing on increasing the WU resources. Hopefully they've got enough spare parts and connectivity on hand to do so, and won't have to rely on buying stuff in.

 

[rise]

FAQ

 

[rise]

 

[VirtualLarry]

Well, I saw this thread, put my secondary boxes (Ryzen R5 1600, already crunching PrimeGrid for the race on the CPU, and GTX 1660ti, was mining) onto F@H on GPU-only (deleted my CPU slot), but were sitting there, waiting for WS Assignment Server or something.

So I started mining back up (they can mine and fold at the same time, amazingly enough, just at reduced performance for each, apparently. Sometimes, not even that much reduced, as mining takes a lot of memory bandwidth (for ETH), and not a lot of compute).

Then I started F@H on my main rig (Ryzen R5 3600, currently also running PrimeGrid on CPU for race, RX 5700 and GTX 1660ti Gaming X), on the GPU(s)-only. Same issue, no WU assignments yet, so I started mining back up on the GPUs, then I got an assignment / WU for the GTX 1660ti, so I stopped that device on the mining, and then I got a WU on the RX 5700, too! So I stopped mining entirely on my main rig.

So far, both the RX 5700, and GTX 1660ti, are both going @ nearly 750K PPD, for a total of 1.5Mil PPD on this rig. We'll see.

Will check the secondary rigs soon, and stop mining on them, too, if they're going ahead with F@H WUs, finally, as well.

Edit: I don't know how much my PrimeGrid CPU computing is limiting my F@H performance. I have BOINC set to use 100% of CPUs, 100% of CPU time, and my PrimeGrid prefs, to allow 3x WUs, with 4x threads per WU. (All three of my rig CPUs are 6C/12T.)

 

[Ken g6]

Hey, I just noticed there's a retry countdown on F@H!



I suppose I should not reboot if that's still counting down.

 

[rise]

Damn, I get ~150k ppd, lol. Every last bit helps I say!

 

[Markfw]

Well, I have 8 of 12 video cards inactive due to the connectivity and server problems.

 

[StefanR5R]

On a positive note, Anonymous contributors had a busy Sunday:



(image credit: EOC)

 

[amrnuke]

Also having issues connecting/downloading. Navi's been sitting here unused for a couple of days now despite restarting, etc.

 

[Markfw]

I am down to 2 of 12 GPU's running. This is pathetic. And its due to errors

11:20:27:ERROR:WU01:FS01:Exception: Could not get an assignment
12:07:25:WU01:FS01:Connecting to 65.254.110.245:8080
12:07:25:WARNING:WU01:FS01:Failed to get assignment from '65.254.110.245:8080': No WUs available for this configuration

 

[borandi]

Somehow another media organization have managed to get a large amount (~1500) in the last couple of days. I wonder if they're scripting some sort of forced communication with the servers

 

[biodoc]

Here's the link to FAH server status.