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

[Malifestro]

Hey team,

I've just now put my 1080 to work. By default it is setup to 'Any disease'. Does that put it to work on the covid work?

 

[Markfw]

Hey team,

I've just now put my 1080 to work. By default it is setup to 'Any disease'. Does that put it to work on the covid work?

You can't specify, but most units are covid-19. Look at the number. There is a list of unit numbers for covid here somewhere.

Edit: my numbers, 8 working, 4 idle (waiting on a unit)

 

[Howdy]

All day no WUs FINALLY snagged one!!!.

 

[mopardude87]

Ah, sucks getting old and forgetful hehe. I just came back for the covid tho I think I might get back into it, I forgot how fun such a nerd can have with this stuff

Yeah its not a bad cause to certainly help out with, not exactly like most of us can do anything but doing this is at least doing something. That and i could stay home and monitor the clients, make sure they just work.

 

[Assimilator1]

If you want to keep you CPU busy with some covid19 beating action, you could switch it to Rosetta.

Haha, "MoparDude" - and Mopar being the current "we'll substitute HP for... everything!" king So your username checks out with your post "no replacement for displacement".

There are some solid x570 boards (Asrock Pro4, for example) with some solid specs for not that much more than a solid B450 board - and it only requires one 8 pin. I just wanted something shinier and honestly got spooked to save some money so dropped back to 16GB of ram and saved ~$30 on the motherboard. So yeah, budgets happen

Mine & the MSI mbrd mopardude87 mentioned have got well reviewed power delivery circuits.
I think the Tomahawk is the one that has the poor audio chip though, IIRC, that's why i didn't go for that one in the end.

God the forum is running slow atm!...

 

[StefanR5R]

Hey team,

I've just now put my 1080 to work. By default it is setup to 'Any disease'. Does that put it to work on the covid work?

Yes, that's correct.

You can't specify, but most units are covid-19. Look at the number. There is a list of unit numbers for covid here somewhere.

The list was posted on page 1 of the thread, posts #18 and #10, but more projects have been added meanwhile:

[FAQ] SARS-CoV-2 (COVID-19) projects and FAH

How do I know that I'm working on COVID-19 related projects?
If you're running a project that's listed here, you're fighting COVID-19 :
CPU : 13862 - 13863 - 14328 - 14329 - 14337 - 14528 to 14531 - 14572 - 14574 - 14576 - 14600 to 14602
GPU : 11741 to 11781

How do I choose to help with the COVID-19 related projects?
To help in the fight against COVID-19, you need to select
- Webcontrol : "Any disease" in the list "I support research fighting"
- Advanced Control/FAHControl : Configure > Advanced, select "Any" in the list "Cause Preference"
The COVID-19 related projects are on top priority and will be assigned automatically.

Spoiler: The complete 'COVID-19 and FAH' FAQ

by toTOW » Sat Mar 14, 2020 10:58 pm


This thread is intended to regroup the Frequently Asked Question about the SARS-CoV-2 (COVID-19) projects. Please read it first.

If you don't find your answer here, the next step is to run a Search on the forum to see if some one else had the same issue before.

1. How do I choose to help with the COVID-19 related projects ?
To help in the fight against COVID-19, you need to select
- Webcontrol : "Any disease" in the list "I support research fighting"
- Advanced Control/FAHControl : Configure > Advanced, select "Any" in the list "Cause Preference"
The COVID-19 related projects are on top priority and will be assigned automatically.

2. Can I choose to work only on the COVID-19 related projects ?
No, you can't.
In FoldingHome, you never choose to work on only one project (or family of projects). The setting described above is a preference, but if no work is available to match your preferences, the system is designed to send you what need to be done at the time of your work request.

3. Is my system able to help ?
There are WU available for both GPUs and CPUs. Folding@Home supports nVidia and AMD GPUs through OpenCL on Windows and Linux.
The CPUs are supported on Windows, Linux and MacOS.

4. My GPU is supported, but it not detected.
You need to update your drivers with the latest ones downloaded from nVidia or AMD website.
Windows Updates often breaks OpenCL support when it updates the drivers automatically. To fix it, reinstall the drivers you downloaded from nVidia or AMD website.
On Linux, in addition to the proprietary drivers from the GPU maker, you may have to manually install "opencl-icd" matching with the drivers you installed.

5. Can you add an option to the list of causes to specifically support COVID-19 ?
No the categories are currently hard coded. We are now in the end of the development of a new client.

6. How do I know that I'm working on COVID-19 related projects ?
If you're running a project that's listed here, you're fighting COVID-19 :
CPU : 13862 - 13863 - 14328 - 14329 - 14337 - 14528 to 14531 - 14572 - 14574 - 14576 - 14600 to 14602
GPU : 11741 to 11781

7. How to fix Web control error ?
For: Google Chrome and some Chromium based browsers,
To fix web control,
Hit f12 on browser to enter dev mode,
right click on refresh button and hard reload.
Also try Incognito mode or clear cache.
Firefox is not affected by this bug.

8. How much does it cost to run FAH on my computer ?
This cost depends on many variables : computer configuration, using CPU and/or GPU, power cost, subscribed power, ...
I sometimes simplify the estimates by saying that 1W used 24/7 for a year will cost about 1€ per year. It could a bit more or less, depending on how much you pay for electricity.

With the following hypothesises (adapt them to your usage/energy provider) :
- regulated power "blue plan" from EDF, with a subscribed power of 3 kVA (0.1546€/kWH in January 2018), the lowest you can get here in France (but with higher kWH rate)
- a computer using 300W when running FAH.

A simple calculation lead to the following estimates :
- running FAH 24/7 will cost a little bit more than 400€ per year, or about 33€ per month.
- if the computer runs only 8 hours per day, the cost goes down to 135€ per year, or about 11€ per month.
- finally, if the computer is run only on the evening (3 hours per day), it would cost ~50€, or a little bit more than 4€ per month.

To be completed ...

 

[StefanR5R]

I am seeing pigs from hell flying South again.
Understandably, given that hell has frozen all over.

Mark_F_Williams ..... 7,537,267 points during the last 24 hours​

xii5ku ....................... 8,115,259​

@Markfw, did I understand correctly that you have 12 GPU slots in operation?
I've got 9 GPU slots. (All are 1080 TI, all Linux.)
My working hypothesis is that during the current workserver bottleneck, the number of folding slots is somewhat more important than the computational throughput of the folding slots.

 

[Markfw]

I am seeing pigs from hell flying South again.
Understandably, given that hell has frozen all over.

Mark_F_Williams ..... 7,537,267 points during the last 24 hours​

xii5ku ....................... 8,115,259​

@Markfw, did I understand correctly that you have 12 GPU slots in operation?
I've got 9 GPU slots. (All are 1080 TI, all Linux.)
My working hypothesis is that during the current workserver bottleneck, the number of folding slots is somewhat more important than the computational throughput of the folding slots.

Yup, all 12 running with the script. 5 2080TI's, 2 1080TI's, 3 2060's, a 2060 super, and a 1070TI. Here is the list: current ppd is 9 mil. 9 of them run linux, mostly 19.2 mint, 2 windows. (the 1070ti and a 1080TI) No idea why you keep beating me. Last 24 hr is 7,712,637

$Edit Here they are:

 

[StefanR5R]

Weather forecast says that the next nights will not go below freezing point. Therefore I am feeling safer now to leave windows open 24hours. And out of curiosity, I copied old foldingathome data directories and executables from older system disks of my dual processor computers over to their current disks. I started them up now:
56 threads slot: 686 k PPD estimated
56 threads slot: 701 k PPD estimated
88 threads slot: 1.03 M PPD estimated
88 threads slot: 761 k PPD estimated

That might keep the pigs flying for a little longer. ;-)

All in all, these four CPU-only computers draw 1.4 kW at the wall for 3.2 M PPD, while the other four computers with 9 GPUs take 2.6 kW for 14 M PPD peak in Folding@home (realistically 7...8 M due to idle times, uploads taking very long etc.) plus ~40 kPPD in Rosetta@home on their CPUs.

(Edit, this means my Pascal-equipped computers are about 2.5 times as power-efficient in Folding@home as my Xeons.)

 

[StefanR5R]

Stats updated, with an ahead/behind table, and with commas!

Thank you!


Day 9: Friday, March 27
14:00 UTC stats
(10 AM EDT, 7 AM PDT)
links: official race stats, TeAm AnandTech at EOC, Tom's Hardware at EOC

Daily production


Our advantage


Overtake estimation

	race points	delta	last 24h	overtake based on last 24h	2-day avg.	overtake based on 2-day avg.	3-day avg.	overtake based on 3-day avg.
TeAm AnandTech	1,074 M​	187 M​	139 M​		119 M​		112 M​
Tom's Hardware	887 M​		121 M​	—​	112 M​	—​	108 M​	—​

Looks good. Keep it up! :-)

 

[blckgrffn]

So, my RX470 (son's PC) and GTX 950 (PC for bike trainer) seem to always have work. It takes them 3-5 hours to chew up a WU but they combine for right around 500K PPD.

The 5700xt which claims about 1.1M PPD when it is really going has work like 50% of the time? I need to pay closer attention to see if there are different GPU tasks for different capacity GPUs.

That said, it's likely to be a placebo thing, since those other slower GPUs are often mid-job when I make my pause-fold rounds I just don't see them idling as much...

PSU shows up today to get this shiny 1080ti in the game

 

[borandi]

PPD on the interface is calculated as

PPD = credits for this unit / time take for the unit

It assumes you have work 100%. It isn't cleverer than that.

 

[StefanR5R]

@blckgrffn, it figures that a slower working slot needs to ask for new work fewer times per day than a faster one.

I don't think larger GPUs receive different work than smaller GPUs. But I actually don't know how the WU schedulers at F@h work.

You could load the logs of your clients into a text editor, then search for the string "Project", and see whether this gives systematically different patterns between your clients.

 

[Ken g6]

Hey team,

I've just now put my 1080 to work. By default it is setup to 'Any disease'. Does that put it to work on the covid work?

You can't specify, but most units are covid-19. Look at the number. There is a list of unit numbers for covid here somewhere.

Edit: my numbers, 8 working, 4 idle (waiting on a unit)

Most but not all current work is COVID. If you're using the advanced control, you can click the project number to learn more about what you're doing.

I just got a WU simulating the human oxytocin receptor!

 

[Icecold]

Is there a BOINC setting that would have it only run certain projects when Folding@home is idle? I've been manually starting GPUGrid when I'm having issues getting Folding@home WU's, but it's hard to stay on top of.

 

[StefanR5R]

You could try

boincmgr -> View -> Advanced View...​

Options -> Exclusive applications...​

There, enter FahCore_22 in "Suspend GPU usage when these applications are running". Verify in taskmanager though whether the precise name of F@h's GPU application is FahCore_22. I took this name from F@h on Linux.

Two caveats: I don't know whether GPUGrid has suitable checkpointing to be interrupted like this. Also keep in mind hat credits decrease in GPUGrid if the turnaround from task download to reporting the result takes more than a day.

(I am routinely using the other field, "Suspend processor and network usage when these applications are running" on my desktop PC to suspend computation in boinc whenever I run system updates and application installations.)

 

[Assimilator1]

Damn it! As if WUs aren't hard enough to come by, I get this


16:19:18:WU01:FS01:0x22:ERROR:exception: Error invoking kernel sortShortList: clEnqueueNDRangeKernel (-5)
16:19:18:WU01:FS01:0x22:Saving result file ..\logfile_01.txt
16:19:18:WU01:FS01:0x22:Saving result file science.log
16:19:18:WU01:FS01:0x22:Folding@home Core Shutdown: BAD_WORK_UNIT
16:19:19:WARNING:WU01:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
16:19:19:WU01:FS01:Sending unit results: id:01 state:SEND error:FAULTY project:11776 run:0 clone:4970 gen:6 core:0x22 unit:0x0000000f287234c95e73c457206b3c34
16:19:19:WU01:FS01:Uploading 8.00KiB to 40.114.52.201

And I think more than once......
Yep

16:25:12:WU02:FS01:0x22:ERROR:exception: Error invoking kernel sortShortList: clEnqueueNDRangeKernel (-5)
16:25:12:WU02:FS01:0x22:Saving result file ..\logfile_01.txt
16:25:12:WU02:FS01:0x22:Saving result file science.log
16:25:12:WU02:FS01:0x22:Folding@home Core Shutdown: BAD_WORK_UNIT
16:25:12:WARNING:WU02:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
16:25:12:WU02:FS01:Sending unit results: id:02 state:SEND error:FAULTY project:14533 run:0 clone:12596 gen:12 core:0x22 unit:0x0000001580fccb025e72f2d73180733b
16:25:12:WU02:FS01:Uploading 8.00KiB to 128.252.203.2

And again


16:26:52:WU00:FS01:0x22:ERROR:exception: Error invoking kernel sortShortList: clEnqueueNDRangeKernel (-5)
16:26:52:WU00:FS01:0x22:Saving result file ..\logfile_01.txt
16:26:53:WU00:FS01:0x22:Saving result file science.log
16:26:53:WU00:FS01:0x22:Folding@home Core Shutdown: BAD_WORK_UNIT
16:26:53:WARNING:WU00:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
16:26:53:WU00:FS01:Sending unit results: id:00 state:SEND error:FAULTY project:11776 run:0 clone:33088 gen:4 core:0x22 unit:0x00000005287234c95e7931368fc0da9c
16:26:53:WU00:FS01:Uploading 8.00KiB to 40.114.52.201

Not had a WU since.
Anyone else getting this?

 

[Icecold]

You could try

boincmgr -> View -> Advanced View...​

Options -> Exclusive applications...​

There, enter FahCore_22 in "Suspend GPU usage when these applications are running". Verify in taskmanager though whether the precise name of F@h's GPU application is FahCore_22. I took this name from F@h on Linux.

Two caveats: I don't know whether GPUGrid has suitable checkpointing to be interrupted like this. Also keep in mind hat credits decrease in GPUGrid if the turnaround from task download to reporting the result takes more than a day.

(I am routinely using the other field, "Suspend processor and network usage when these applications are running" on my desktop PC to suspend computation in boinc whenever I run system updates and application installations.)

Thank you! That works as expected on the two Linux Mint machines I have running. The Windows one did not, and I did verify the process name/file path of the EXE, but I didn't spend much time looking into it further. It probably would be good timing just to move that one over to Linux Mint anyways.

 

[borandi]

Looks like Linus is actually spinning up a 100 TiB SSD + 10 Gbit server for FAH.

Folding@home Server Stats

 

[ZipSpeed]

So, my RX470 (son's PC) and GTX 950 (PC for bike trainer) seem to always have work. It takes them 3-5 hours to chew up a WU but they combine for right around 500K PPD.

The 5700xt which claims about 1.1M PPD when it is really going has work like 50% of the time? I need to pay closer attention to see if there are different GPU tasks for different capacity GPUs.

That said, it's likely to be a placebo thing, since those other slower GPUs are often mid-job when I make my pause-fold rounds I just don't see them idling as much...

PSU shows up today to get this shiny 1080ti in the game

I noticed the same thing. My 1060 has been getting consistent work these last few days, but my 2080 Ti has barely done anything these last few days.

 

[Markfw]

I noticed the same thing. My 1060 has been getting consistent work these last few days, but my 2080 Ti has barely done anything these last few days.

Try putting "client-type" = "beta" in the expert part of the config setup.

 

[Assimilator1]

Looks like Linus is actually spinning up a 100 TiB SSD + 10 Gbit server for FAH.

Folding@home Server Stats

Nice
But why is the assignment rate zero for that 1 along with 2 of the azure servers?.......... has 'No CS', whatever that means ( no explanations there)

 

[ZipSpeed]

Try putting "client-type" = "beta" in the expert part of the config setup.

Thanks, will try this later.

 

[StefanR5R]

Looks like Linus is actually spinning up a 100 TiB SSD + 10 Gbit server for FAH.

Folding@home Server Stats

Nice
But why is the assignment rate zero for that 1 along with 2 of the azure servers?

Presumably only a limited number of WUs was put on the servers at first when they join production.

.......... has 'No CS', whatever that means ( no explanations there)

This means this work server is not paired with a "collection server". The latter would back the work server up for the upload of client results. (I have seen many occasions in the logs when the client tried unsuccessfully to upload to the work server, retries failed too, and then the upload went through to the collection server.)

More explanation of the serverstats page is in this foldingforum thread:
Server Status - What it means

 

[Ladicek]

What is more effective for my old computer rosseta@home or F@h
I have 2500k @3.3 Ghz, AMD Radeon 460 RX
or both rosseta for CPU, F@h for GPU ?