I looked for scorch marks when it happened but the board looks pristine.
Device ID : 32
Device Revision : 1
Firmware Revision : 2.01
IPMI Version : 2.0
Manufacturer ID : 2623
Manufacturer Name : ASUSTek Computer Inc.
Product ID : 4499 (0x1193)
Product Name : Unknown (0x1193)
Device Available : yes
Provides Device SDRs : yes
Additional Device Support :
Sensor Device
SDR Repository Device
SEL Device
FRU Inventory Device
IPMB Event Receiver
IPMB Event Generator
Chassis Device
Aux Firmware Rev Info :
0x15
0x00
0x00
0x00
% ipmitool -I lanplus -H 192.168.0.172 -U admin -P admin sel info
Get SEL Info command failed: Unknown (0x18)
% ipmitool -I lanplus -H 192.168.0.172 -U admin -P admin sensor
Error obtaining SDR info: Unknown (0x18)
Unable to open SDR for reading
% ipmitool -I lanplus -H 192.168.0.172 -U admin -P admin power status
Get Chassis Power Status failed: Unknown (0x18)
It was the power supply, I wired up one I had for another project and it powered on, though now I have the infamous 92 code. Progress at least. Just hope the psu didnât take out my 6000 Ada when it imploded.
That is interesting since you didnât get code 92 the first time. Almost makes me wonder if the PSU load regulation or voltage ripple affects the memory training.
What PSU were you originally running? Besides the magic smoke release it sounds like it might be a good unit.
a silverstone hella 1300, the one I have in now Is the same model but only 1200, but should still be enough juice?
It might not be a question of how much juice but how âcleanâ the juice is going into the motherboard.
I just took a look at the test for each power supply and the 1200R definitely has bad ripple compared to the 1300R; the actual voltage regulation was fine though. The 12V rail is probably the most important since thatâs what feeds memory and PCIe.
1200R Ripple on 115V:
1300R Ripple on 115V:
Wow, thatâs interesting. Though would be kinda wild if the tolerances are that tight, but kinda makes sense with what we have seen from other builds so far.
If youâve got a 230v outlet you can use, trying it might be able to help assuming voltage ripple really is the source of memory training problems (big if).
For some strange reason the voltage ripple is significantly better (but still not great) on that same power supply when using 230v.
And IT LIVES, plugged it into a 230v and it powered up. I had to reset some things due to attempting to reset the cmos but yeh everything is working. This board seems SUPER sensitive to voltage ripple.
Now to go shopping to try to find another 1300.
That is very interesting, Iâm going to mention this in the other threadripper code 92 thread.
There are PSUs that have even better ripple than the 1300R too. Itâs still sort of a pain to find the good ones but atleast cybenetics posts figures for all the PSUs they test.
I am a little disappointed to see that the Noctua cooler is trying to blow, or suck, horizontally rather than vertically, AND that the cooler only has one tower versus my NH-D15 that has two. How are the temps on that thing?
They were mentioned earlier:
Yeh, the fan isnât super great and kinda noisy. Hope we get a solid AIO solution for this socket at some point.
I have kinda-confirmation of this in a different way â if +5 SB shuts down on a PSU the symtoms are similar. So if it gets to where it wont turn on again, its worth turning it off and unplugging it. +5v SB overheats after about 30 mins on one PSU I have but I can still turn the machine on? but it doesnt actually come on I just get click click
Sorry, I must have missed that part.
Hmm, thatâs what was happening on the original power supply. It would go click click but only the BMC/ILOM would actually turn on. Though I let it sit over night before I gave up and tried another power supply. So I donât think it was an overheat situation.
Did you let it sit without the cord plugged in (or PSU switch turned off) though? The +5 SB will continue to run and possibly overheat even when the computer is off.
It was unplugged over night. Honestly was about to take a mallet to everything to release some stress and go buy a xeon but thankfully I had calmed down enough over night to try the other power supply after it still did not come up in the morning.
It almost sounds like this could be two separate PSU issues kind of running in parallel (to make the troubleshooting more interesting).
The 12v ripple affecting the memory training and the +5 SB overheating/breaking and the system booting up weird/not at all because of it.
There donât seem to be many PSUs with high +5 SB rails. The Silverstone Strider 1500 can do 6A on the +5 SB but it is pretty old.