tictoc

Sorry for my late start team. Internet went out in my area Friday night, and then my router decided to begin a slow death sometime during the outage. This turned into a nightmare to troubleshoot, because I was just assuming that the internet was still down. I restarted my router a few times, once it looked like the internet was back up, and that is when I started to suspect that, in a rather unbelievable coincidence, my router had also decided to die. My modem just serves as a dumb bridge to my EdgeRouter, so I went through a bunch of config/re-config on the router and modem, before I noticed that the only port that was lit up on my router was the console port. Hooked my laptop up to the modem, took it out of bridge mode, and lo and behold I had internet access. Currently using my ISP modem as my router, so things are back up for now, but I will probably have some additional down-time once I get my new router. I should be able to time it with the start of a WU, so hopefully it won't affect my folding output.

Double post because I'm the captain, and that's what captains do. Adding the TFA logo to the OP.

@Avacado @axipher Are we all ready to start our reign of dominance? I'll be folding a 980 KPE for at least the first month. I ran through some testing on a few different GPUs, but the 980 was the best compromise I could make between actual folding output and ETF points. I have a few GPUs that might be better for the comp, but the actual PPD was just too low for me to justify running 24/7, when I have other gear sitting on a shelf with respectable output.

This is a great topic. Having mostly ran AMD GPUs for the past decade, I'm pretty used to the notion of undervolting+overclocking (especially in the last few generations) to achieve equal performance at lower power draw. Efficiency is king, especially for systems that are going to run 24/7.

Unless you want it, and then you're free to roll your own keys and encryption. The AMA from the source is more or less worthless unless you were looking for an advertisement for Windows "cloud" services. It will be interesting to see how this evolves. Looks like jailbreaking Windows will be a thing in the future. It will also be interesting to see how in place upgrades work. I guess mbr2gpt will now be a part of the Windows installer. Outside of this being the first step for more intrusive and restrictive DRM, I don't really see the point of secure boot and TPM being a requirement. If anyone has it, I'd love to see a link to the non-marketing, actual engineering reasons behind these requirements. "It makes your device more secure" has zero substance. The minimum CPU requirements are even more absurd. If there is so much focus on helping me secure my device, then why is Microsoft even letting me run an 8700k? What kind of "testing" are they doing that would make such an arbitrary minimum CPU requirement?

I am probably not going to be much help on Windows. Do the GPUs show up in device manager at all? Only suggestion I would have is to just add GPUs one by one to see if it is an issue with that many GPUs. Also, you could disable HD Audio and some USB ports in the BIOS, to check if it is an allocation issue.

I know that I had some shenanigans when I tried to run headless with that board. Are you able to boot into the OS? If not you might try putting a GPU with display out in the primary GPU slot. Also, make sure your boot drive is GPT formatted and not MBR, since you have to disable CSM in order to enable Above 4G decoding.

I know that option is in there somewhere, but I don't remember for sure where it is. I think it's in the "Boot" section of the UEFI. The last time I ran it I had multiple GPUs running plus at least one GPU on a m.2->PCIe card.

The trend in HPC, especially with high density muti-GPU nodes, is water-cooling all the things. 6U systems like this dual Epyc w/ 8x NVIDIA GPUs are becoming much more common. https://bizon-tech.com/bizon-zx9000.html

Most of my gear that runs 24/7 is water-cooled. As long as you follow some best practices like EPDM tubing, distilled+Mayhems biocide and inhibitor, dual pumps in serial, there is not really anything to worry about or any additional maintenance. I haven't touched my workstation in 16 months other than to top off the fluid one time, and it runs 24/7 @ about 1900-2000W.

That is looking great, and so is your F@H production. Awesome to see all that hardware running a volunteer compute project.

No Caselabs, just a cheap clone. Thermaltake Core x5.

A few pics of the stripped down case before hardware starts going in.

I ran through all the 4.x ROCm releases, and ROCm 4.0 is definitely the one to stick with for F@H.

https://www.formulamod.net/products/barrowch-fbehd-01-external-expansion-display-8-8-inch-high-resolution-lcd-screen-for-pc-cpu-hardware-temperature-monitor

Good luck. I'm rooting for this build to become a reality. If Windows keeps giving you headaches with driver and GPU detection, you might think about running Linux. I haven't messed with the the Maxwell Teslas, but Keplar Teslas were pretty painless. If you decide to give that a go, I can definitely give you a hand.

Sorry for the double. I posted about this in the Foldaholics thread last month, but then forgot all about it when I fired up some additional GPUs for this months race. This probably won't apply to too many people, but if you are running AMD GPUs in Linux with ROCm, do not update to ROCm 4.2 if you are folding. There is a pretty massive regression (-50% or more) on most WUs. Right now I'm running an older closed source driver, but I am going to so some testing with ROCm 4.0 which had really good performance on F@H when it was released. **Edit** I'm now back on ROCm 4.0, and the performance is great. OpenCl via ROCm 4.0 does not have the performance regression that I saw with ROCm 4.2. ROCm OpenCl 4.0 is about 8% better than the old closed source driver on the WUs that I am currently running (p18108).

Nice work. Having a proper board (I'm guessing an X299 Sage or similar) will make your life much easier in the long run.

@BWG Do you need me to fill out your form? No changes on my end, and I'll let you know what hardware I'll be folding on a few days before we start.

Yeah, your going to get me on the next update.

Got it, blame Canada. Although, UTC is "the real time", so I'm with @axipher

Thanks. I didn't see the count down. I won't be putting up a ton of points, but I'll get some things folding.

When does this actually start??? The OP has some whacky days and times. Is it 2021-07-22 @ 1 AM UTC?

^This 100% There is pretty much no way that Valve can bargain bin it. It is commodity hardware running on an open source OS. Unless they drop in some proprietary driver for the controls, anyone who buys one has more or less a guarantee that it will work unless Valve abandons Linux altogether, and even if that were to happen, there's nothing that's been revealed that would preclude someone from running games via other means (native Linux games, emulators, Wine, installing Windows, etc) Everything that Valve has done to improve gaming on Linux is open source: Proton, DXVK, ACO, RADV. On top of that it is running on an AMD GPU, which has a very good kernel driver + mesa. I don't personally have a use for this, but outside of some KickStarter/Indiegogo products, it is a pretty unique device.