NVME crazy high temps >100 deg under load - any ideas / thoughts?

Hi - monitoring is showing me that I am hitting over 100deg on my main NVME on all 4 nodes on my cluster. Any suggestions like a special NVME cooler that people have seen?

Its this one - Lexar NM790 4TB SSD with Heatsink, M.2 2280 PCIe Gen4x4 NVMe 1.4 Internal SSD, Up to 7400MB/s Read, Up to 6500MB/s Write, Internal Solid State Drive for PS5, PC, Laptop, Gamers (LNM790X004T-RN9NG)

Ta

Chris

I’m not sure if you can use a NVMEcooler as the board kind of has them built in. I suggest to get some additional air flow on the board first. check out where your main drive is located and ensure there is airflow to that space.

Those are pretty high temps. Did you test your SSD separately? In another system or with an NVMe enclosure? Is it all good like that, and is it problematic in the FW Desktop?

My temps were a bit high too compared to what I like to see, but I reduced them with about 10-20 °C by installing the front 80mm fan. Honestly, I think the 80mm fan should be mandatory. Because the heat from the APU comes straight on the main SSD.

But your temps are way too high; double-check if the heatspreader is all good too, or replace it with something better if needed.

Thanks both! I think I need to completely revisit cooling for the mini racks. Both near the boards and NVME and also for the whole rack. I will see what cooler other than the stock NVME one might fit. I also ready that it can make sense to have the CPU fans permanently at 100%

The EC controls the fans, and the EC only reads temps for, ambient, cpu, gpu. It does not read NVME or RAM temps so you probably just need a tool the reads all of those and controls the fans better.
I think a user has a made a tool that already does that.

Thanks all. Having dug into this a little more I am swapping out the heat sink on the NVMe for something beefier, correcting the orientations of my CPU and network card fans (who knew you needed to fire air at the heatsink :slight_smile: ) and adding fans to the bottom and top of the case.

Chris

The default fan settings for the desktop seem to be built with the APU’s health in mind, this ignores the fact that the power supply and SSD’s can’t handle these temperatures. This is probably my biggest frustration with the device as a whole since if your OS ends up forcing you to use the default bios settings for the fan controls instead of the ones I personally set, as then you strictly need a third party tool to get thermal performance that preserves the SSD’s and keeps the power supply fan from kicking into emergency mode constantly.

I had room temps that were far below normal room temps last week due to an AC error (Like 15-16c,) and the power supplies fan was still practically constantly kicking into overdrive when the deskop was placed on a vented metal rack designed to help electronics stay cool even when it was only on for like. One. Hour.

Honestly this isn’t my first run in with a default fan setting feeling for the sake of being quiet where they instead keep the fans at slow speed where they rattle instead of sounding like a fan, and then some other device in the computer has a much worse sounding fan constantly kicking into emergency mode because the main fans running at like 150-230 rpm until the CPU hits 70C then it finally decides to turn on.

Its more annoying here because software for the framework desktops fan controller is relatively new, and the setups can involve extra annoying steps compared to a lot of the more conventional fan controllers.

All agreed. I have ordered a Linux compatible fan controller and some big fans. I will mount the PSUs outside the rack as given they are using extenders they have the reach. All that with the nvme cooler should put me in a better place.

Chris