Unfortunately, I don’t remember and the 2600X was just a loan from an aquaintance until I got my 3700X so I cannot test it again.
The issue however seems to regard the reading the motherbord’s various temperature sensors (none available on my two units with 3.04 and 3700Xs), including ECC memory. In Windows I could get TDie via HWiNFO. This is fixed in 3.09.
Honestly so far I’ve heard nothing from either companies’s Support Team that they really know where the CPU_PROCHOT issue is coming from.
I don’t know if this is just corporate BS (“We never make any mistakes and any issues must be coming from how you use our product!”) and behind the scenes they might be scrambling to fix this.
Either way, I detest such communication practices since ultimately the truth will most likely come out and I also would like other things to have more priority in my spare time than me collecting different PSU models to test them with these pretentious little Micro-ATX boards
When all is set and done the Optane drive is running at PCIe 3.0 x4 in PCIe slots 4, 5 and 6, at PCIe 2.0 x4 in M2_1 and at PCIe 3.0 x2 in M2_2.
The file names should make the content of the images obvious, I didn’t make screenshots of the CrystalDiskMark results since its GUI had a weird bug (scaling?) that made it look ugly AF. I saved each configuration’s test results to a coresponding .txt file.
All tests were done with the same Optane unit since I couldn’t free up my second one.
huh so it worked for you in the M2_2 slot?!
Did you have to set anything for that in the bios?
I got the original intel U2 to M2 adapter and I got nothing.
Which bios are you running?
Asrock send me the 3.09 version you mentioned, so I’ll try that tomorrow.
But as I understand you recommend to put it in slot 4 or 6?
Yes, everything is working in this regard (for me) with 3.09.
The only change I made is the PCIe Switch for PCIe slot 4, no special stuff for M.2 set so default settings.
Yes, in PCIe slots 4, 5 and 6 the Optane drive (and any other NVMe SSD) will operate at its maximum performance as you can check in the .txt files attached to my previous posting.
I activated Above 4G decoding support since it sounds nice to hsve support for anything so maybe try that if the update to 3.09 isn’t solving the issue for you. Maybe it has nothing to do with that? Since I’m a normie I can’t tell.
The tests were done with a Seasonic Focus Plus Platinum 550 W PSU since it is the only one I got that doesn’t trigger anything weird in the IPMI log.
Have you tried just connecting power to the PSU (don’t power on the system), wait until IPMI is ready (about 1 min, a green LED is blinking then), log in via management LAN, update the BIOS (just get 3.09 from ASRock Rack’s support, ignore the more buggy previous versions), then reset everything to factory defaults via the IPMI web interface?
Just to be sure: Have you checked the SATA power connector on the U.2 end of the Intel OEM M.2-to-U.2 adapter?
I got two units and one adapter’s build quality is a little worse than the other one’s and most PSUs’ SATA power connectors are a pretty loose fit on it.
Did your Optane unit with the bundled adapter work in any other system?
If that doesn’t help I’d try the other M.2 slot (performance between PCIe 3.0 x2 and 2.0 x4 is almost identical). If it still doesn’t work I’d try to get a PCIe-to-M.2 adapter card like the Delock card I linked to in my test run posting that is a pretty good value (not cheap, it just works reliably at PCIe 3.0 x4).
After my new finding that the rather well-built Seasonic PRIME Ultra Titanium 850 also shows some CPU_PROCHOT IPMI log entries, I received an email from ASRock Rack support in which they hint at me to wait “a few days” for the next BIOS release (above 3.09).
If I switch from my Captain Obvious to the little dusty Captain Subtext goggles I think that is corporate speak for “We found out that we did a whoopsie but can’t admit it.”
At least I hope so so that this issue is finally concluded.
When I initially got my first X470D4U I didn’t frequently check the IPMI log - after about a day of testing I noticed that sometimes the sysmen’s performance was extremely (really extremely) slow and that the CPU (2600X) was only running at 550 MHz.
Since then I can almost predict when it’s triggered depending on what of the three PSU models I have on hand is used.