Folding Forum

The 2080ti runs hot under full load but shouldn't be running that hot tbh and might be the suspect for the crashes - airflow in case is probably lacking my guess is that internal case temp may be in the low 40s at best from the ssd/cpu temps - and possibly in 50s when cpu loads up as well as the gpu - this would spike the gpu temps even higher? ... Is the AIO fan/rad mounted inside another radiator? - as per ajm it the photos looks a tad odd and don't clarify quite the setup ... a bit of investigation as to what is causing the card to run that hot (lack of airflow in case, fans not working, dust in heatsink, even bad thermal paste) is probably worth it ... some folders reduce power to gpu to cool/make it more efficient - might be worth trying this (although already not max wattage) ... any chance of some more photos showing the big picture of the airflow?

Water cooling in a pre-built computer? That's unusual.

Water cooling is mostly for overclocking and having those two features generally requires more vendor support than the guy who did your pre-building is willing to provide.

I'm making a lot of (unwarranted?) assumptions here, but FAH stresses that computer's overclocking in ways that the builder didn't expect. Traditional overclocking benchmarking software can be insufficient ... but he doesn't know that.

There are now quite a few vendors who offer very decent water-cooled (AIO or custom) prebuilt systems.
https://www.letsbld.com/
https://www.ibuypower.com/gaming-pcs/si ... id-cooling
https://www.originpc.com/gaming/desktops/
https://www.cyberpowerpc.com/
Even some best of: https://www.tomshardware.com/reviews/be ... ,5198.html

But do they set the overclocking strictly for games or do any of them consider people who run FAH?

I don't know how they set up the overclocking, but I'd guess they don't take that much risk with prebuilt customers.
At any rate, a water-cooling in a prebuilt should suggest that the system will be able to handle FAH better than most, shouldn't it.
Here, the water-cooled CPU looks good, only the air-cooled GPU (not overclocked) seems to hinder the use of FAH.

Where do you see problems with air-cooled GPU (not overclocked) prebuilt systems?

Are they supported by the builder?

The only one I've seen was in a cabinet with no space for the hot air to escape.

I don't, generally. We just happen to have such a problem here. This thread is about this problem. 5 pages and still no solution.

https://imgur.com/a/bhbcSCO

The case is a 275R as you've seen before, since you provided insight about the Power Supply lane being accessible on the reverse side https://www.corsair.com/us/en/carbide-275r-case . I've thought about removing the face panel. There is half an inch of clearance there for what I assume to be intake? There's about 4-5 inches of clearance in the rear, and the top is unencumbered except for a (magnetic) mesh layer.

Every fan, GPU and case, all seem to be spinning without issue, but; Shouldn't case fans be increasing up to ~5000+ RPM as well? Even after a day of cranking, the case fans haven't notably exceeded 2000, even though the GPU fan has ramped up to 5000+

https://i.imgur.com/qd0VpG9.png

This also isn't a complete list, seeing as there are 3 on front, 2 on top, 1 at the rear. Perhaps it just doesn't know about Radiator sensors? Or sensors for those don't exist? Even discounting the 2 radiator intake fans, there's still 4 including the 1 front fan above the radiator that also blows inward.

All the fans do appear to be blowing the right way , the 3 fronts are intake, the other 3 are output. I haven't sat down and looked to see if this is a 3 vs 4 pin header limitation from the motherboard or… something. I am very much in the deep end in all this, coming back in the hardware, plus PC world, after more than a decade of being out of it.

The build is basically okay. But the placing of the radiator on the lower part of the intake at the front produces a lot of hot air blowing directly into the Graphics card. It would have been much better to place the rad at the top, as exhaust (but this case only accept a 240 at the top), or at least on the upper side of the intake, so that the Graphics card gets a little bit of fresh air.
I would test the system with the side panel off, to see how much it helps.

You quite probably don't have any fan capable of 5000 RPM in this PC, it is a collective value (several fans connected to the same port). There is no "radiator sensor". On a custom loop, you can add sensors on the loop to monitor the temperature and flow, but AIO usually don't come with such niceties.

Thoughts:

1) (If possible - not sure if your radiator is a 240 or not but the pictures make me think it may well be) Move the radiator and mount at top above existing top fans so they push air through it and out - and if this won't fit then mount on front but higher up and you need to invert it so the pipes are at the bottom of the rad - the way it is might cause issues if the coolant runs low as the pump is taking connecting at the top of the rad (check youtube for the debates as to why this is/is not good) - the top mount will have the significant benefits of both taking the cpu heat straight out of the case and also allowing a much better airflow from front to back for cooling the gpu ... It may be a tight fit and need care when fitting but top mount will be significantly better tbh.

2) Investigate why case fans don't seem to be spinning up - either use the system fan controller software or grab one of the free fan control software packages and play with increasing base fan speed and curves to see if you can get the case ambient down - but you do need to get control (if you haven't already - you may find that just removing the radiator impediment to the front fans might make a big difference to this ambient anyways.

3) Adjust the fan curve for the gpu up a bit so that it cools gpu better/earlier - it may just be a bit conservative and not be responding quick enough to heat spikes.

... and then hopefully the gpu temps can be controlled down and the crashes become a thing of the past !!

Ahoy, long time no see. I'm posting this a bit over-optimistically, but I have a reasonable amount of confidence in it;

Sometime in April a new behavior manifested, my system was shutting down entirely, completely randomly, immediate power off (power light on the power button turns off too) and it turned back on after about 2-3 seconds. Note: NOT a Blue Screen, I have "automatic restart" for those turned off and there is no intermediate screen, just an immediate spin down, power off, everything. It pretty well shellacked my ability to do anything persistent (gaming, work, etc.). There were a bunch of anecdotal things that led me to rule out the power supply, the biggest one being that I have "Automatically start up on power restore" disabled in the EFI, which should have meant that if there were a power problem, the system would not automatically turn back on. But it did, so I didn't think it was power source related, but some motherboard bug or something.

After days of teeth gnashing, many many more days of memory testing, and eventually sending my PC off to the OEM that built it for diagnosis, I got it back with two new sticks of memory, and … the shutdown behavior happened within the first few minutes of me plugging it back in and getting going with it. Frustrated with everything but thinking of the symptoms, I ordered a new power supply, set everything back up today, and my computer has been up for 3 hours, which is the longest it has been on uninterrupted in over a month.

After everything initially checked out, naturally my first test was turning Folding on and seeing if it survived while I went back to work for a bit longer. Come back, still running, great! On a lark, I also re-enabled the use of CUDA ( viewtopic.php?p=345237#p345237 from Page 3 in this thread) which in the past was an immediate crash. It wasn't the shutdown behavior, but negative behavior nevertheless. I undid that change and am writing this post with Folding running full tilt.

So, it appears that all of this teeth gnashing, drama, and nonsense, was probably my previous power supply dying a slow death from the start. Being unable to draw a full power rate, and eventually not even being able to draw enough for standard/moderate use. (Fun fact: It even shut off / restarted during the EFI boot phase on at least one occasion. Though it never did during memtest IIRC, may or may not be related/relevant.) I ordered another nearly identical PSU, slightly different Cooler Master model number, one letter and one digit off, installed it today, and here I am writing this post.

Knock on wood that it's not continued random badness and it still falls over at some point, but even being able to write this post, with the combination of Folding with CUDA cranking away, has me very optimistic.

I appreciate everyone sticking with me and helping me along in the first phases of the badness. This was a doozy before it was just a straight mess, and I do believe the two were related.