Question TROUBLESHOOTING: MSI Pro Z790-P Wifi Red CPU Light

[kdt_bowie]

OK, so here is my system inventory:

• MSI Pro Z790-P Wifi
• Core i9-14900K
• 128GB RAM
• MAG A750GL PCIE5 power supply
• Gigabyte Radeon RX 9060 XT video card (added 2 weeks ago)
• 3 NVMe disks (2 linux, 1 windows)
• Lian Li 3-fan liquid cooler
• (don't know the case make/model, threw box away)

Machine is about 7 months old
Linux used mainly for development/research
Windows used mainly for X-Plane12
Don't do any crazy over-clocking or anything like that

I added the GPU about 2 weeks ago specifically for the sim. I asked the guys at the store if my 750W power supply was enough for all this and they said it was plenty.
Also recently added Logitech USB yoke/throttle/rudder pedals for the sim, but I can't see that as relevant

Everything was fine up until about 2 days ago. I had played for several hours in the sim, and shut the machine down late that night. The next afternoon ... power on, no boot, no display, and the red CPU light on and solid.
The next morning it booted. I played in the sim for about 15 mins. It was a little choppy at first but then settled down. I went to download Open Hardware Monitor, but as I clicked the link to download (not install, just download the installer), the machine instantly hard-powered off (as though I had pulled the power cord out of the wall), and now gives me the on-and-solid red CPU light ever since
The cooler is off of the CPU for the moment. When I turn the machine on, the CPU stays cold, and I only leave it on for a moment or two.

I have tried:

• pulling the GPU card
• pulling and reseating the CPU, no bent pins that I could see. (the CPU was a little goopy with paste, cleaned that up
• pulling and reseating the memory
• pulling and reseating the ATX cable to the MB
• adding another 8-pin CPU power cable from PSU to the MB (only had one previously, was told that's all I need)

My current thought, since the last thing added was the GPU, and that I have been gaming in (what I assume to be) a fairly demanding app, was that I overloaded the PS. All of the voltages on the CPU connectors seem to be correct, though ... 12.12V.
Next most likely, the MB has failed
Next, the CPU itself

 

[COLGeek]

Do you have access to another PSU to swap for testing?

 

[kdt_bowie]

Do you have access to another PSU to swap for testing?

No, I'm afraid not. Maybe time to invest in some hot spares, LOL

 

[DavidM012]

So would it go over if I say: 'does this sound like a logical hypothesis to you: The gpu produced a transient pwer spike that overloaded the psu, and zapped the vrm which causes the cpu light to activate even if the problem may now reside in the cpu power deliver system - the motherboards vrms not the cpu itself.

Therefore one could venture switch mobo and psu to something like a corsair rm-e or x 850w. While the mag psu is tier a, a search suggests it's lower tier A and while a transient spike wouldn't be expected to trip it's oc we're just in if buts and maybe territory here' one has to prove a hypothesis, no? So the prescription would be switch mobo AND psu first. Psu only = red cpu light due to mobo vrms and more frustration. switching PSU + mobo could = frustration if the CPU is indeed fried also. ' Is there a more concise way to float this?

 

[DavidM012]

a lightening strike or power surge? Any recent inclement weather ?

 

[kdt_bowie]

a lightening strike or power surge? Any recent inclement weather ?

No, nothing like that

 

[kdt_bowie]

So would it go over if I say: 'does this sound like a logical hypothesis to you: The gpu produced a transient pwer spike that overloaded the psu, and zapped the vrm which causes the cpu light to activate even if the problem may now reside in the cpu power deliver system - the motherboards vrms not the cpu itself.

Therefore one could venture switch mobo and psu to something like a corsair rm-e or x 850w. While the mag psu is tier a, a search suggests it's lower tier A and while a transient spike wouldn't be expected to trip it's oc we're just in if buts and maybe territory here' one has to prove a hypothesis, no? So the prescription would be switch mobo AND psu first. Psu only = red cpu light due to mobo vrms and more frustration. switching PSU + mobo could = frustration if the CPU is indeed fried also. ' Is there a more concise way to float this?

I have a Corsair RM850e on the way here today, I'll try tomorrow first thing.

 

[DavidM012]

*Reminder to remove all the modular power supply cables of the old psu and use only the cables supplied with the corsair with new the corsair psu - just ticking the box maybe you know that already

 

[kdt_bowie]

*Reminder to remove all the modular power supply cables of the old psu and use only the cables supplied with the corsair with new the corsair psu - just ticking the box maybe you know that already

Good call, will do

 

[kdt_bowie]

New PSU did not fix it ... I guess MB next? Then CPU? I REALLY don't want to dig that MB out of the case LOL

 

[DavidM012]

Well you have a choice - cpu or mobo, or maybe find a cheaper cpu variant and try it in the mobo.

I don't think there's any upside one way or the other - if you get the cheaper cpu variant and try it, but the mobo's dead it's just made the mobo replacement more expensive. If you switch the mobo and the cpu doesn't fire up it's getting clearer. Could be a shorter route unless the cpu isn't working. It's a coin flip.

Dunno if you could ask any pals who could test the cpu in their rig for you.


You could try clearing the mobo cmos - switch off the mains take the cmos battery out and press the pwr button to drain capacitors - let it sit for 5 mins then put the cmos battery back in.

Do you still have a previous gpu around? Maybe try with a known good gpu that wasn't involved in the event just to rule it out before sploshing on a mobo.

There's also ram to consider. It was in the system when things went haywire. You could try pulling one stick or the other and trying them one at a time. Doesn't tell you anything if all the ram sticks also bought it.

Then there's also the choice - it could just be easier to take to a repair shop. long as they're honest, who knows? If you're comfortable proceeding this way with testing the parts then you save their hourly rate or what parts they try to push on you.

Dunno if they do computer clubs or anything like that these days where you could try different parts.