Jump to content

Folding Rigs suddenly getting very low points for about a week, unable to resolve. Looking for help!

Hi All,

 

I am participating in the October FAH competition and I have been struggling for close to a week with low PPD #s that come on after weeks of solid PPD and have not corrected themselves in time or regardless of new jobsr.

 

My stats and some history.

 

Rig 1 - 8x GPUs, i7 6700k - Was getting 1.8 - 2.2 mil PPD, now topping out around 600k PPD

Rig 2 - 2x GPUs, celeron g3930 - Was getting a stable 600k PPD, now topping out around 15k PPD

Rig 3 - 1x GPU, Ryzen 2700x

 

 

I had a 10 GPU mining rig. I installed a 6700k here to add more cores, and removed two GPUs and added them to a secondary miner chassis that has a G3930. This should be providing 1 thread per GPU on each device. While I do not expect this to be optimal, I have been getting relatively stable speeds (600k from the 2GPU system, 2mil-ish from the 8 GPU system). I then have a third rig (my gaming system) with a 5700 XT that is running along just fine (no issues, ranging 800k - 1.1 mil PPD). From my reading, I expect each RX 580 to get 300-450k PPD, but I have not be able to verify this outside of anecdotal reports via forum posts.

 

Here is what has happened: Rig #1 (The 8x GPU system) and even worse, Rig #2 (the 2x GPU system) are crawling and have been for days. The 2x rig is barely getting 15k PPD (this is only like 1% of what I would expect from 2x 580s) and the 8x rig is getting around 600k (much better, but I've been getting around 2-2.5 million for weeks now regardless of job. I leave them both on "Any" in regards to type of work they are allowed to do, so I know my PPD will fluctuate, but they seem to have 'crashed' now regardless of job.

 

I have tried canceling/deleting jobs, forcing them to only do specific job types, rebooting, messing with page file and RAM amounts, try running only 1-4 GPUs (to free up more CPU cores) and verify if PPD-per-GPU goes up (it did not). CPU on both systems are never maxed out, and I believe that while the optimal setup is 1 core per GPU, CPU usage seems well within the limits on both systems. I'm sure that CPU is bottlenecking my overall MAX PPD, but up until now I have been getting relatively stable numbers that were much higher than what I have been seeing the last week.

 

There only seems to be a few posts from people over the years regarding low PPD, so I've found relatively little that I can do. The main thing people bring up is running a username + PPD, which I am and have verified is in place on all 3 rigs. I'm at wits end about what to do, and reaching out for any assistance that the community can provide. I started here as I feel the LTT FAH forums may in fact be the most active FAH community on the net.

 

Thanks for your time.

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

are you by any chance running PCIe X1 riser cables on your mining rigs? 

in mining, the PCI bandwith isn't all that important, but in F@H you need a PCI connection with lots of bandwith since it's talking to the CPU all the time.

 

also, did your points crash right around the new CUDA update by any chance?

 

Link to comment
Share on other sites

Link to post
Share on other sites

Not surprised if your first two rigs had unstable PPD. The whole CPU is dedicated to the GPU's, and nothing is left for the O/S, in terms of available threads. Each GPU requires 1 CPU thread. It's also recommended to have at least 4 pci lanes per GPU.

Current Build

 

Link to comment
Share on other sites

Link to post
Share on other sites

OK this is all good info, thanks. I am running 1x risers and unfortunately am stuck with it for now (as this is the only time of year these machines will be folding).

 

What doesn't make sense to me is why the one rig is only getting (celeron + 2gpu) <30k points. The CPU is not maxed out, and starts at 500-600k for the first few days. Last year, I ran this same processor with 8 GPUs and averaged around 100k per GPU. Now I'm barely getting 15k per GPU on this 2 rig system. I'm contemplating putting the 2 cards back into my 8 rig system and returning it to 10 GPUs, as I was getting much higher PPD on these 2 cards in that system.

 

Any other ideas on what could be wrong or if there is any additional info I can provide? I know I will not get any where near my maximal PPD per card with this config, but I also know I should be getting way higher than 15k PPD.

 

Thanks for your time!

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×