Folding Forum

I saw for sale locally CMP 50HX cards, based on the TU102 chip. TPU says performance is between 2080 and 2080S. Can I assume they are natively supported by fah through the nvidia driver? The no video output situation doesn't bother me as I have a small Quadro card for video output.

joescrubbs wrote: ↑Sat Feb 11, 2023 12:18 am If I run on a 8-GPU NVIDIA DGX, does FAH software do anything smart to coordinate among multiple workers, or do I just start 8 FAH workers?Thanks!

What's your folding username, I would like to follow up on the evolution. Thanks!

May I recommend you get at least a 1060 3GB or one of those P100-106 GPU without video outputs designed for mining exclusively? You can keep the ti for display purposes only and harness the power of a mining GPU. Indeed you cannot select the size of the project. You might also want to update the dri...

My GTX 1050 is doing as much as 360 kPPD while taking 67W. My 2070S is doing 3 mil PPD with 210W.

2070 S is doing 3 mil PPD plus on Windows these days, for those wondering, with driver 512.xx

So, one year later looks like a 1080Ti would be a better investment than a T4 if the cost of electricity is not an issue. But in general, how much is a T4 producing these days?

Tesla T4G proves to be a Quadro RTX 4000 built on the TU104 chip. Techpowerup says that performance wise, an RTX 2070 SUPER is 16% faster at gaming, and my HP 2070S pulls consistently over 3 mil PPD, on Windows, and you can look up my stats on EOC as it is the only card folding rn.

I jumped on the occasion and got a 2070 Super, these days is doing around 3 mil ppd.

Hi there. I am thinking of buying a Titan Xp or a 1080ti, which 99% of the time will be on folding duty. In case I don't find one at reasonable pricing, what other cards from the rtx 20/30 series have similar production numbers? Efficiency is not an issue, the amount of electricity spent is not an i...

Just do a VM with Linux.

It has been suggested that the optimal smp client setup to have 24 cores allocated, no more. As for the 1080ti there is some fluctuation over time, look here. https://folding.lar.systems/gpu_ppd/brands/nvidia/folding_profile/gp102_geforce_gtx_1080_ti_11380 Seems you could get better results by insta...

@bourbonbelly so, who is the candidate?

Thread necro : so, if you get a machine with 56 threads, what would be an optimal setup? Two clients with 24 threads each?

Yes that's what I meant. Proper pcie bandwidth not really : I had 290x folding on a 1x slot and it lost 2-3% of the ppd.

Historically, the launch drivers are really optimized for gaming and nothing else. A few months down the road the drivers will offer more ppd as they mature and they also get optimized for computing as well.