Join the MicroCenterOfficial Folding@Home Team: #257944

PowerSpec_MichaelB

@MightyMayfield Do you guys have any secret tips for keeping your GPU's working? I feel like we have to restart our system every other hour or so in order to get new work units, or else half our GPU's sit ready for a long time. Our system has an estimated PPD of nearly 10M, but we get a fraction of that at the end of the day due to this phantom inefficiency that prevents our GPU's from getting work units until rebooting.

We've reinstalled the OS and client several times, but I don't know if it's a configuration issue or not. I've not seen any difference with the advanced flags enabled either. I've considered simply scheduling Windows to restart every 1-2 hours and see if that helps long-term.

Yellowbillycat

TSMichaelB said:

@MightyMayfield Do you guys have any secret tips for keeping your GPU's working? ...

@TSMichaelB I haven't been doing anything special.  Two Linux system and one Windows, and I haven't seen any problems with getting GPU WUs.  Though GPU WUs seem to be a lot harder to come by Friday evening through Sunday, sometimes.
Only one 8 core cpu doing folding, and that is being fed 24/7.

MightyMayfield

TSMichaelB said:

@MightyMayfield Do you guys have any secret tips for keeping your GPU's working? I feel like we have to restart our system every other hour or so in order to get new work units, or else half our GPU's sit ready for a long time. Our system has an estimated PPD of nearly 10M, but we get a fraction of that at the end of the day due to this phantom inefficiency that prevents our GPU's from getting work units until rebooting.

We've reinstalled the OS and client several times, but I don't know if it's a configuration issue or not. I've not seen any difference with the advanced flags enabled either. I've considered simply scheduling Windows to restart every 1-2 hours and see if that helps long-term.

Yes! Without giving away too much of the automation secret sauce, if you pause a work slot and then resume it, it will reconnect to the download server in 2 minutes instead of the previously allotted time.

PowerSpec_MichaelB

MightyMayfield said:

TSMichaelB said:

@MightyMayfield Do you guys have any secret tips for keeping your GPU's working? I feel like we have to restart our system every other hour or so in order to get new work units, or else half our GPU's sit ready for a long time. Our system has an estimated PPD of nearly 10M, but we get a fraction of that at the end of the day due to this phantom inefficiency that prevents our GPU's from getting work units until rebooting.

We've reinstalled the OS and client several times, but I don't know if it's a configuration issue or not. I've not seen any difference with the advanced flags enabled either. I've considered simply scheduling Windows to restart every 1-2 hours and see if that helps long-term.

Yes! Without giving away too much of the automation secret sauce, if you pause a work slot and then resume it, it will reconnect to the download server in 2 minutes instead of the previously allotted time.

Thank you, that's extremely helpful to know. Now to figure out a way to do it with 4 GPU's and 5 clusters of 20 cores without it becoming tedious every time, lol. I think some scripting to grab some pixel colors might be required.

MightyMayfield

TSMichaelB said:

Thank you, that's extremely helpful to know. Now to figure out a way to do it with 4 GPU's and 5 clusters of 20 cores without it becoming tedious every time, lol. I think some scripting to grab some pixel colors might be required.

It is doable! I believe in you!  
Though I didn't use color recognition. I imagine that would work as well.

1TM

1TM

JADO

ChrisP said:

Mild gaming GPUs have way more folding@home power than expensive CPUs do.  Check it out - I'm folding with two 3950X systems and a 2700X system.  Folding with 80 cores *sounds* really impressive and it definitely heats up the room, but the reality is a single GTX1060 or 5500XT video card can fold twice as fast as all those CPUs combined!  If you want bang for the buck, fold with your video cards!!

True Dat....My 6 card RX580 rig produced over 175 work units and over 5 million points the first week and that was with significant down time due to startup and shakedown issues...….  When I built it for mining Etherium it cost about $4,500.  Today you could probably build it about $2K.

wuffy68 said:

Tramp78 said:

wuffy68 said:

Where was everyone when Folding@home Cancer, Alzheimer's, Parkinsons, Ebola, Diabetes and Influenza needed help with research? 

I just heard about this yesterday. Should I stop?

Ryau is absolutely correct ... COVID19 is Folding@home's number 1 priority right now ... but don't be surprised when they run out of COVID19 work units, you will also be receiving Cancer, Alzheimer, and Ebola work units as well. Curecoin merely rewards folding@home participants who register on their system no matter what work units are received. Not asking anyone to switch teams - just trying to highlight how there are some innovative teams on Folding@Home that rarely get credit.

JADO

ChrisP said:

Mild gaming GPUs have way more folding@home power than expensive CPUs do.  Check it out - I'm folding with two 3950X systems and a 2700X system.  Folding with 80 cores *sounds* really impressive and it definitely heats up the room, but the reality is a single GTX1060 or 5500XT video card can fold twice as fast as all those CPUs combined!  If you want bang for the buck, fold with your video cards!!

True Dat....My 6 card RX580 rig produced over 175 work units and over 5 million points the first week and that was with significant down time due to startup and shakedown issues...….  When I built it for mining Etherium it cost about $4,500.  Today you could probably build it about $2K.

wuffy68 said:

Tramp78 said:

wuffy68 said:

Where was everyone when Folding@home Cancer, Alzheimer's, Parkinsons, Ebola, Diabetes and Influenza needed help with research? 

I just heard about this yesterday. Should I stop?

Ryau is absolutely correct ... COVID19 is Folding@home's number 1 priority right now ... but don't be surprised when they run out of COVID19 work units, you will also be receiving Cancer, Alzheimer, and Ebola work units as well. Curecoin merely rewards folding@home participants who register on their system no matter what work units are received. Not asking anyone to switch teams - just trying to highlight how there are some innovative teams on Folding@Home that rarely get credit.

kmiller922

Just a heads up everyone who has put out a notice of a new outbreak not just for covid but ebola and some others as well. The other point I was going to make is even though we selected covid-19 as a priority the system will substitute with the next high priority item in the cue.  I hope that we can keep as many possible to stay the course till a cure is found for all of the research that's needs to continue to be done.

kmiller922

@MightyMayfield man are you ever crushing it. Guessing you found the warp drive now congrats on top 900.
Heck i was shocked to see i'm in the top 20k. LOL

MightyMayfield

kmiller922 said:

@MightyMayfield man are you ever crushing it. Guessing you found the warp drive now congrats on top 900.
Heck i was shocked to see i'm in the top 20k. LOL

Working on some cool stuff to hopefully increase output again. Always more science to be done! Microcenter is doing great as a team. We're almost to the top 100!
Edit: MicrocenterOfficial has been losing team members, though

teamtempest

First work unit I've noticed that isn't related to Covid-19 - not even any specific disease ("potassium ion channel inactivation and conduction").
My monthly statistics did not reset for May. Instead they just keep building on last month's results, showing the team at #47 for several days now. Five billion points, but not overtaking or losing position to nearby teams. Is this a problem with my web control?

MightyMayfield

Getting a lot of upload errors recently. Not sure if our firewall is causing issues or if it is FAHservers. Many 0 byte returned errors. Hm..

kmiller922

MightyMayfield said:

Getting a lot of upload errors recently. Not sure if our firewall is causing issues or if it is FAHservers. Many 0 byte returned errors. Hm..

Might be the server I had that issue the other day took 35 attemps to send the packet.

kmiller922

For some that are intrested in some updates on the project here's a like to one of the talks about the moonshot project for covid-19.  

https://www.youtube.com/watch?v=e8frs67HW78&feature=youtu.be

kmiller922

Ladies and Gentlemen I'd like to congratulate everyone on the work so far to get into the top 100 teams. Let's keep up the great work.  

ryau

kmiller922 said:

Ladies and Gentlemen I'd like to congratulate everyone on the work so far to get into the top 100 teams. Let's keep up the great work.  

Yes! Well done everyone!
https://stats.foldingathome.org/teams

ChrisHarshman

kmiller922 said:

Ladies and Gentlemen I'd like to congratulate everyone on the work so far to get into the top 100 teams. Let's keep up the great work.  

Awesome!  Keep on folding. 

cine_chris

kmiller922 said:

Just a heads up everyone who has put out a notice of a new outbreak not just for covid but ebola and some others as well. The other point I was going to make is even though we selected covid-19 as a priority the system will substitute with the next high priority item in the cue.  I hope that we can keep as many possible to stay the course till a cure is found for all of the research that's needs to continue to be done.

@kmiller922 I'm a recent F@Her.  Yes, Folding is tedious, akin to watching paint dry or grass grow and not w/o frustrations like the WUs that were choking my 1660s.  As I'm located in the BDD (basement data dungeon), the upcoming summer temps will likely force me to throttle my F@H efforts, but these challenges have me thinking in terms of managing energy costs and looking at a Fall strategy as fancy new, higher efficiency gear begins to arrive. Little Mental Games that keep me interested in the Folding effort, like chasing the Thread-Ripper demons.  

kmiller922

cine_chris said:

@kmiller922 I'm a recent F@Her.  Yes, Folding is tedious, akin to watching paint dry or grass grow and not w/o frustrations like the WUs that were choking my 1660s.  As I'm located in the BDD (basement data dungeon), the upcoming summer temps will likely force me to throttle my F@H efforts, but these challenges have me thinking in terms of managing energy costs and looking at a Fall strategy as fancy new, higher efficiency gear begins to arrive. Little Mental Games that keep me interested in the Folding effort, like chasing the Thread-Ripper demons.  

I totally understand about temps and energy cost believe me. One of the first thing that i have been looking at over the past number of years of building my own systems is energy efficiency. What shocked me the most on energy cost was when i upgraded my monitors, when flat screens first came out i was seeing a reduction of energy usage by almost 30% by the time I changed all monitors and tv's in my house. The other thing I'm looking into now is there a energy savings going to water cooling versus fan as well as more efficient components to my builds. 

The biggest thing to me personally is the side effects that are now coming up that are effecting our children and a first for me now is grand children, For myself paying a bit more to assist in the research to save anyone with these issues is still and will be as long as i can possibly be able to be more than worth every cent that is put into it. That's just me being who I am, being in the fire and EMS field that's just the way the I am. Besides that I'm use to watching the grass grow LOL, I also work at a golf course.

teamtempest

Visited the suburban Minneapolis store today and watched the store's build work at F@H for a bit. Very pretty machine, but kind of strange in operation. Lots of F@H CPU threads going, maybe a half dozen or so, all apparently multitasking on a single physical CPU. The powerful GPU glowering in the case was doing nothing. My own machine isn't very powerful, but there's never any time the CPU and the GPU aren't both working unless I've set F@H to minimum power - then the GPU stops whatever it's doing until I turn it up again.

cine_chris

Wommwalsh said:

Ahh-wooooooo! (lone wolf howl from Minnesota)

My main ‘point’s getter’ is a i9-9980XE / GTX2060S whose parts came from the St. Louis Park store. It must do its day job – but is folding nights and weekends.

My previous ‘mothballed’ workstation (a 10-year-old Dell T7400 with a pair of Xeon E5430’s and GTX660Ti), a spare laptop (i7-6700HQ / GTX 970M), and an extra RX570 are running full time.

I started on April 3rd when I received the MC email. Mrs. Ahh-wooooooo just saw the electric bill from last month (+$55) and is lobbying to turn one of them off.

@Yellowbillycat (another lone-wolf?) is chasing the store-teams too.  The weather is starting to heat up in the ATL area, so I'll have to start throttling back on the F@H effort soon.

cine_chris

@Wommwalsh Those old Xeons are power hogs, but in Minnesota you can use the heat anyway, right?  I'm using a mix of old re-purposed gear & newer stuff. I have a couple of old towers that I've stripped down to bare 'open-frame' chassis that I'm tinkering with & culling open-box GPUs from the local Marietta,GA MicroCenter. 

cine_chris

teamtempest said:

Visited the suburban Minneapolis store today and watched the store's build work at F@H for a bit. Very pretty machine, but kind of strange in operation. Lots of F@H CPU threads going, maybe a half dozen or so, all apparently multitasking on a single physical CPU. The powerful GPU glowering in the case was doing nothing. My own machine isn't very powerful, but there's never any time the CPU and the GPU aren't both working unless I've set F@H to minimum power - then the GPU stops whatever it's doing until I turn it up again.

@teamtempest If you check out other forums (like EVGA) you see that they usually relegate the CPU to managing the box and feeding multi-GPUs.  I do have a couple of CPUs folding but they are efficient multi-core processors with robust heatsinks too.  CPUs just aren't as efficient as GPUs for this task.  I have a Ryzen 5 teamed with a MSI 2060 Super that is showing 2MilPPD (just checked) the Ryzen is producing a mere 90K of that 2 million PPD, but the combo works well.  I should point out that the cost of that combo you could probably build 3x systems for 1x of the sTRX4 ThreadRipper systems (just a WAG guess).  I have a 2070 super on an older system that is only slightly faster than the 2060, so IMO, architecture affects the total system efficiency.  Tinkering with the gear efficiency is interesting to me.  

kmiller922

Now how cool would it be to get a hold of that new Nvidia AI chip and put that to work if it was affordable. There i go dreaming again.

MightyMayfield

teamtempest said:

Visited the suburban Minneapolis store today and watched the store's build work at F@H for a bit. Very pretty machine, but kind of strange in operation. Lots of F@H CPU threads going, maybe a half dozen or so, all apparently multitasking on a single physical CPU. The powerful GPU glowering in the case was doing nothing. My own machine isn't very powerful, but there's never any time the CPU and the GPU aren't both working unless I've set F@H to minimum power - then the GPU stops whatever it's doing until I turn it up again.

Its possible the WU wasn't currently active on their GPU. If not setup properly (or servers are down......) things do idle a lot as many of us have discovered.

Kirito

I've been running three Lenovo ThinkStations (1 C20x with single processor, 1 with two processors, and a C30 with two processors.  I had to repaste the dual CPU C20x, which has been my main machine (until recently replaced by the C30) for over six years.  Temps had been running around 82 F for awhile.  After repaste they came down to 75 F.  admittedly I am using all the RAM slots so air flow remains a problem.  My C30 now runs around 68 - 70 F at 100 % capacity.  Fewer Ram slots though it has more RAM.  Still looking to get a heatsink for the single processor C20x, the one that slides under the CD-ROM.  Would also have to get more RAM. 

Yes the Xeon's are power junkies.  My electric went up fifty a month and I only added the C30 recently.  Worth every penny.  Don't have the money for GPU's though the ThinkStations are set up to run Tesla K40's out of the box.  Just not in the budget.  If you happen to have one you find antiquated by your standards please feel free to donate.

Matter of fact, that might be a good idea.  Donating GPU's to help all us lower tier members to produce more.  A coordinator, a database of systems online for the project (Those who have been here more than six months, of which I wouldn't even qualify) and people willing to donate GPU that are compatible with the FAH program, OS and and system board.  maybe a cheat sheet on how to run CPU and GPU at the same time.  Maybe MC could throw in old GPU's that are open box / refurb as prizes for individuals who have achieved high levels with little resources.  Just a thought.  Having the technology, especially as fast as you guys change things up, could make a difference in getting more people into FAH and on our team as well.  While I agree we should all be doing this out of the goodness of our heart, using our ample skills and resources, some people just cannot give as much as they could.  But they could do more with a little help.

I'll let you guys kick that around awhile. 

Kirito

kmiller922

Ok so you can say I come up with some insane ideas at times but thought I would run a thought I have past all of you for your thoughts.  I've recently came accross adds about external gpu. Now would there be away using the same type of thunderbolt port set up to add more gpu to a system other than just one? What would be the best way to go about it? Or is there a way to use the home network and use it like network storage is done? I know weird thought lol

cine_chris

I've been using data from a couple of docs that are kept current (sorry, don't recall the source).  Most interesting to me was the total cost factor PPD/Cost + PPD/Watt.  
https://docs.google.com/spreadsheets/d/e/2PACX-1vRcR50eJc1MC60klR5d8R1jeYspg3RAMbeEeguvybakdmy0kM3E-_IDcjbsUwhzhGrzhgSpPRqEQ7Ek/pubhtml
and
https://docs.google.com/spreadsheets/d/1vcVoSVtamcoGj5sFfvKF_XlvuviWWveJIg_iZ8U2bf0/pub?output=html#