random system hang

[PantherX]

... know that Nvidia offers an interface to connect 2 cards to make them work together, but have not tried it. Would FAH see them as 2 slots if tied together?..

If you use SLI, F@H doesn't use it and will see two independent GPUs which it can use.

[bruce]

Right. Don't bother.

The features that let two cards work as one are ignored by FAH. It does improve the frame-rate of the display (if you're a gamer) by distributing the rasterization process across alternating GPUs but FAH will still treat them as independent devices.

[djgibbons]

Nvidia asked me to use only the K4200 and disable the RTX 4000 to see what happens. The computer never froze overnight, even with FAH running at Medium. This morning I reactivated the RTX 4000, but left both monitors plugged into the K4200. One interesting detail is that the K4200 is running about 4C cooler, and the monitor cables are really warm. I think the cables are cooling the K4200 just enough for stability. If the current setup works, I will then move one monitor back to the RTX 4000 and see what happens.

[PantherX]

While I haven't heard of monitor cables used to dissipate heat, if it works, who am I to challenge the PC Gods

[djgibbons]

What I found out is that when the RTX 4000 is active as well, the K4200 slowly starts to warm up. I saw pixels flickering on my 1K screen, and a short while later my computer froze. The temperature of the K4200 was back up by 4C. This tells me that the K4200 is getting hot enough to malfunction. I will figure out a way to cool it more, so that it can still be active.

[JimboPalmer]

You should be able to manually increase the fan speed, and or reduce the GPU clock speed to reduce heat. I use MSI afterburner

https://www.msi.com/Landing/afterburner

[djgibbons]

LOVE THIS! I downloaded MSI Afterburner, and the immediate impact is ~20C drop on both cards (I made a custom curve and shifter the MAX point over to 65C. I will leave it for a while, but later might try FAH on Full setting to see what happens. The immediate effect makes me think that fan control on the cards is wanting. I will ask Nvidia to see what they think.

[PantherX]

Ah, yes MSI Afterburner is awesome and I have always set my fan profile to run at 100% 24/7 and the automatic overclock just works its magic.

[djgibbons]

How is the automatic overclock enabled?

[PantherX]

When it comes to Nvidia GPUs, they use a feature called GPU Boost (https://www.nvidia.com/en-us/geforce/te ... echnology/) which is considered overclocking. Thus, I used the "automatic overclock" but do note that you can use the OC Scanner (https://www.techspot.com/news/78031-nvi ... lable.html) to get more performance but folding is more demanding than gaming so I let GPU Boost work. Stock frequency on my GPU is 1607 and it boosts up-to 1900 to 2025 depending on WU and ambient temperature.

[djgibbons]

It does not seem to work on Quadro cards. But my computer ran overnight with FAH on Full setting, so I am happy with that.

[toTOW]

So to summarize, the issue was a lack of cooling of one GPU ?

[djgibbons]

Well, that was short-lived. Even running a lot cooler, the K4200 started to freeze the system again, after a day and a half. At one point I had one monitor on each card, and on the last freeze-up the K4200 monitor was purple. Very indicative of which one has issues.

It almost seems like there is a build-up of data or temp files, and then it runs out of room and freezes. I need to contact Nvidia again. Right now I have the K4200 disabled.

[toTOW]

In my opinion, the K4200 is dead ... and since it's an old Kepler that is very likely the next generation to be removed from driver support, there's no point investing time and money on it ...

[djgibbons]

True. It still works to support a monitor, and I might hang on to it just in case a friend needs a replacement. I have contacted Nvidia again to see if they can make sense of it.