Folding Community Board

[Favebook]

5 minutes ago, Bitter said:

Got the 1650! Popped it in, FAH picked it up but wouldn't fold on it, had to set openCL to 0 and Cuda to 1 manually. Took a few minuted but grabbed some work and hammered my CPU because i3, I need to swap in the i5 I guess. Overall PPD took a huge nose dive with both cards running, to about 500K PPD vs the lone P106-100 at 700K PPD. Currently trying to get JUST the 1650 to fold but fahcontrol is being wonky and doesn't seem to like to issue commands down to the fah process properly. It works fine from my laptop remotely, but at the actual PC it's being weird which is new in the past few days. Anyway, finally got the P106 paused and the 1650 folding by itself at 75W is doing a quickly estimated 600-700K PPD and barely making any noise to do so. I really think I'm going to grab a 2nd 1650 of the same cheap OEM style and just run one or two of those for the long term folding box. Now to procure an i5 S lower power CPU, I gave my 4590S away to my dad's PC build.

Disable CPU folding if you plan to use 2 GPUs on low-tier CPU.

Delete all slots. And then add first GPU>OK>add second GPU>OK>Save (do not edit gpu-index, opencl-index or cuda-index, just leave them all at -1), you will have much higher PPD, less heat (due to no CPU folding) and both of your cards will work without any problems IF you have correct drivers and good enough PSU.

[Bitter]

1 hour ago, Favebook said:

Disable CPU folding if you plan to use 2 GPUs on low-tier CPU.

Delete all slots. And then add first GPU>OK>add second GPU>OK>Save (do not edit gpu-index, opencl-index or cuda-index, just leave them all at -1), you will have much higher PPD, less heat (due to no CPU folding) and both of your cards will work without any problems IF you have correct drivers and good enough PSU.

Already not CPU folding, the issue is that Nvidia drivers hold one CPU thread per GPU when it's in use by F@H so both physical cores of the i3 are held for the GPU leaving nothing for the system overhead to run with except the fictional hyper-threading 'cores'. Running the rest of the system takes away enough resources from the Nvidia driver threads that it slows folding down substantially. I have a 4590 or something in the basement but I'd prefer a low power i5 for long term use, I'm seeing about sniping a T series 4 core i5 but prices are very meh at the moment, they're niche CPU's.

 

PSU is a 600W Gold Apevia that's a decent re-brand of some thing else. I've had the issue with cards not picking up work until I manually set the Cuda or OpenCL before, usually after a reboot it'll fix itself but not this time for some reason, running the latest or maybe now the 2nd latest nvidia drivers. OS overhead could get a little lighter if I wasn't so GUI dependent still, but Xubuntu is fairly light. I could probably disable some more things like CUPS and other non necessary services, and I will at some point. There's just no getting around Nvidia wanting a dedicated thread for each of their GPU in the system. AMD doesn't need that, but the RX580 is too power hungry and hot.

 

Edit: Was having some kind of weird issue with the P106 not wanting to stop doing work even though it showed as paused in fahcontrol, so I pulled the card out. I'll fold with just the 1650 for a while, probably going to sell the P106-100 at some point I guess.

 

Edit2: Yeah, fahcontrol was really acting up funky, it wasn't actually changing the information shown when I was selecting different slots. 300-350K PPD on the 1650 currently, maybe it's a weak WU. I'll let it run for a couple days and see what's up.  Ah I checked the logs, it's running OpenCL not Cuda for some reason now. I'll fix that tomorrow or else I'll be up all dang night messing with this. Once it's sorted out it should be a better producer.

[justpoet]

10 hours ago, MrFinn741 said:

Do you have single or multi CPU systems? When testing my dual Xeon server I only got 75-90% utilization when running all threads in one slot, probably because the CPU-CPU communication can't keep up. I did try more slots and fewer cores per slot but same low util, probably have had to force which cores a slot should use. Switched over to Boinc instead where the cores run individually and got 100% util no problem.

Both.  Both are also running Mac OS X and getting 100% utilization either way on most WUs.

[Bitter]

So I ended up staying up trying to fix it, broke it worse, and found out that the 16GB SSD (Don't ask...) was full while trying to reinstall the nvidia drivers because Cuda wasn't working, and ended up cloning the drive to a spare 128GB Samsung 830 SSD and it's still not fixed but I am too tired to continue working on this so I'll pick up tomorrow evening. Don't even know if this fixes it, may just do a clean install and start from scratch because a clean canvas can be easier than trying to paint over your old lines.

[cbigfoot]

5 hours ago, Bitter said:

So I ended up staying up trying to fix it, broke it worse, and found out that the 16GB SSD (Don't ask...) was full while trying to reinstall the nvidia drivers because Cuda wasn't working, and ended up cloning the drive to a spare 128GB Samsung 830 SSD and it's still not fixed but I am too tired to continue working on this so I'll pick up tomorrow evening. Don't even know if this fixes it, may just do a clean install and start from scratch because a clean canvas can be easier than trying to paint over your old lines.

make sure to update the F@H Client.  7.6.21 haven't had any issues if its 7.6.1x or some of earlier versions cant remember the number had issues with multi gpu.

[Bitter]

2 hours ago, cbigfoot said:

make sure to update the F@H Client.  7.6.21 haven't had any issues if its 7.6.1x or some of earlier versions cant remember the number had issues with multi gpu.

Installed new versions of both this morning. Now folding after some other screwery...with Cuda correctly. May try adding the P106 back later, I want to figure out some better fans for it so it's not screaming all the time.

[Bitter]

On 4/26/2020 at 4:17 PM, Gorgon said:

All you need is CoolBits 12 (8 + 4) for Fan Control and Over-clocking. For multiple GPUs I usually use:

sudo nvidia-xconfig -a --cool-bits=12 --allow-empty-initial-configuration

Bringing this back from the dead, tried that, black screen, had to login tty and delete xorg.conf to get gui back. Suggestions?

[Gorgon]

22 minutes ago, Bitter said:

Bringing this back from the dead, tried that, black screen, had to login tty and delete xorg.conf to get gui back. Suggestions?

edit /etc/X11/xorg.conf adding

Option "Coolbits" "12"

to each "Screen" Section.

 

service gdm3 restart

or reboot

[Bitter]

11 hours ago, Gorgon said:

edit /etc/X11/xorg.conf adding

Option "Coolbits" "12"

to each "Screen" Section.

 

 

service gdm3 restart

or reboot

I'll give it a whirl tonight, I'm running the screen from the Intel iGPU, does that make a difference? Also, I don't currently have an xorg.conf. Should I let Nvidia X-Server settings produce one and then edit that file or should I have x-config produce one? What's so odd is that I had this working with the P106, but it's not working with the 1650. I reinstalled the Nvidia drivers to get the 1650 folding with Cuda after I removed the P106 from the system.

[Bitter]

22 hours ago, Gorgon said:

edit /etc/X11/xorg.conf adding

Option "Coolbits" "12"

to each "Screen" Section.

 

 

service gdm3 restart

or reboot

Scored a "35W" i5 4590T for the system so I'm going to pop that in and start from scratch with both cards installed, sometime next week probably, I need to bodge some better fans on the P106, it's just too loud as it is now. The 1650 is folding away at around 450K PPD which is a little lower than I had hoped but for how quiet it is and how little power it uses (averaging around 70W) I really can't complain. When I start from scratch with both cards and the new CPU I should have an easier time with things because I won't be messing with trying to make things work after the fact. I hope. Still don't know why coolbits keeps giving me so much trouble when I had it working previously.

[Bitter]

Got FoldyPuffSr online for the weekend to leapfrog me a few places up and since it's cold the heat won't bother me. Approx 5500 PPW for the Vega56 and approx 6400 PPW for the 1650, efficient little thing just want moooorrreeeee from it. I also need to take into account the rest of the system power draw, FoldyPuffSr is not an efficient system for long term use, it idles around 40W from the CPU when folding and is powering 32GB of RAM as well, it all adds up over time. I'd like to replace the P106-100 with a 1660 or something like it in the next year or so for the sake of better returns per watt as well. Ampere would be nice to buy into but it's $$$, maybe when they release the low end Ampere.

[Critaek]

Am i the only one who is not getting any covid moonshot WU? I got two few days ago and that's it.

[cbigfoot]

11 hours ago, Critaek said:

Am i the only one who is not getting any covid moonshot WU? I got two few days ago and that's it.

all im getting on my 3 gpus 2x2060s and 1x5700xt are moonshots.

[Bitter]

I'm running a 13428 on my 1650 and a 14906 on my Vega 56.

[GOTSpectrum]

  

2 minutes ago, piratemonkey said:

the only thing with that is I've tried, but you've never opened my PMs

Thought we should move the conversation here, If you PM'd me and I didn't open it, it is probable that I didn't see the message as I get MANY messages on the run-up to an event.

 

It is also possible that I did view it and simply forgot to reply. Either way, I'm sorry you got overlooked, I'd say I will remember next year but it's unlikely I will. 

[piratemonkey]

Just now, GOTSpectrum said:

  

Thought we should move the conversation here, If you PM'd me and I didn't open it, it is probable that I didn't see the message as I get MANY messages on the run-up to an event.

 

It is also possible that I did view it and simply forgot to reply. Either way, I'm sorry you got overlooked, I'd say I will remember next year but it's unlikely I will. 

solid decision. In fairness to you, I think the time I did pm you was the covid event, and I was very new so you wouldn't have recognized me. 

I'll be sure to PM you as soon as I hear any rumblings of the event next year  

and thank you again

[GOTSpectrum]

Just now, piratemonkey said:

solid decision. In fairness to you, I think the time I did pm you was the covid event, and I was very new so you wouldn't have recognized me. 

I'll be sure to PM you as soon as I hear any rumblings of the event next year  

and thank you again

And thank you for folding!

 

I'll look forward to having you on the team next year! You pay want to follow either myself, this thread or both to make sure you don't miss it! 

[Parideboy]

Event is over and my 390 decided to start digivolving into a 6800. Seems fair.

[Bitter]

Thankgoodness it's over, maybe I can climb back up the board a bit again without the Vega56 going.

[TVwazhere]

Starting January 1st, 2021, we will be handing out awards based on new higher folding badge thresholds. These new values have been decided due to the increase of power a GPU can provide; this has made it much easier to achieve badges in recent years. Anyone holding a current badge will be grandfathered in! This means you won't have your badge downgraded after the new thresholds come into effect. 

 

F@H Badge Thresholds

In order to be rewarded with the Folding @ Home badge, please note the following requirements. (subject to change)

• Contributor - When you have contributed more than 21 WUs 5 Million Points, you will qualify for this badge, to show your commitment to LTT and the team. 
• Bronze Contributor - When you have contributed more than 1 Million 25 Million points, you will qualify for this badge. This reflects your serious support of the team. 
• Silver Contributor - When you have contributed more than 25 Million 100 Million points, you will qualify for this badge. This reflects your serious and continued support of the team.
• Gold Contributor - When you have contributed more than 100 Million 500 Million points, you will qualify for this badge. This reflects your dedication and continued support of the team and your very generous donation of resources.
• Platinum Contributor - When you have contributed more than 500 Million 1 Billion points, you will qualify for this badge. This reflects your dedication and continued support of the team and your very generous donation of resources. You are now a legend! 
• Diamond Contributor - When you have contributed more than 1 Billion 2.5 billion points, you will qualify for this badge. This reflects your insane dedication and continued support of the team and your extreme donation of resources. Tales of your folding adventures will be whispered through the ages; you are now mythical.

If you wish to have your current points reflected in your badge, please submit your request before Jan 1st 2021.

 

[vsteel]

Wow, guess folding month is really over.  I didn't realize what an affect it had on the overall team output.

 

[marknd59]

53 minutes ago, vsteel said:

Wow, guess folding month is really over.  I didn't realize what an affect it had on the overall team output.

@Den-Fihad to drop his power bill back down

[Guest]

1 minute ago, marknd59 said:

@Den-Fihad to drop his power bill back down

It was so ridiculous that they sent someone over to do a manual reading because they thought something was broken.

[TVwazhere]

1 hour ago, vsteel said:

Wow, guess folding month is really over.  I didn't realize what an affect it had on the overall team output.

 

The graph makes the sudden loss of the F@H Event look more dramatic than it really is. In reality, the points drop was only 12%, remember that bottom line is still 2.6 BILLION PPD, not 0.

 

At our current rate, we will pass curecoin in about a year and be the #1 team aside from Default team (Which most of us don't consider to be a real team)

[GOTSpectrum]

13 minutes ago, TVwazhere said:

(Which most of us don't consider to be a real team)

THEY ARE NOT REAL!!!!