dying CPU/MOBO?

[maddangerous]

I've got a 7950x and MSI x670E godlike

BIOS is version 7D68v16.

I went to reboot my computer today, and was met with a black screen before POST.
I disconnected power, held the power button for 30 seconds to drain caps, and was able to POST and boot.

If I reboot, or shutdown the computer, I have to disconnect power and drain the caps again.

I can get into the BIOS, if I try to change a memory related setting, the BIOS locks up


The only setting I've changed, which was about 5 days ago now, was MSI's high-Efficiency mode was set to enabled. I was able to disable it but it didn't make a difference..

 

No OC on anything except the GPU, and it's been 100% stable until today

Any thoughts? Hardware bad?

I haven't gotten too far into troubleshooting yet due to lack of time.

Edited October 25, 2023 by maddangerous
spelling

[maddangerous]

Forgot to include that during a couple of the reboots, I recieved a "New CPU installed" message, for whatever that's worth

[UltraMega]

You need to do more troubleshooting on your own to narrow it down more before we can help you. Rule out as many parts as you can (try one ram stick at a time, unplug all drives except boot drive, etc). 

[J7SC_Orion]

...hard to diagnose long distance without more info. One thing I would try no matter what in such a situation is to unplug the PSU completely and then take the mobo battery out. That should result in a factory fresh reset (you might also want to get a spare mobo battery, just in case). Also, does your board have a dual bios ?

[maddangerous]

2 hours ago, UltraMega said:

You need to do more troubleshooting on your own to narrow it down more before we can help you. Rule out as many parts as you can (try one ram stick at a time, unplug all drives except boot drive, etc). 

 

Very true, just don't have a lot of time at the moment (we just had a baby recently). 

I feel like the new cpu message isn't a good sign though, honestly. the only time i've seen that in my experience, was failing hardware. Certainly doesn't mean that's always the case though.

 

I'm going to at least pull RAM, and shut off all of the M.2 drives except boot (all the drives are under M.2 shields built on the motherboard, but I can disable them in the EFI). I don't have any SATA devices.

 

Board has a dual-bios, but I have to pull some things to get to the switch.

 

1 hour ago, J7SC_Orion said:

...hard to diagnose long distance without more info. One thing I would try no matter what in such a situation is to unplug the PSU completely and then take the mobo battery out. That should result in a factory fresh reset (you might also want to get a spare mobo battery, just in case). Also, does your board have a dual bios ?

 

So, while I haven't pulled the CMOS battery, there is a reset to defaults type of button on the I/O shield, I just tried that a few minutes ago to no avail.

 

 

It's just so weird, I've had 0 problems this entire time. Built the rig on February-ish, and I never enabled AMP due to the BIOS bugs. when AGESA 1.0.0.7c came out, after the fix, I enabled AMP. Checked voltages in hwmon64, everything looked fine.

Edited October 26, 2023 by maddangerous

[UltraMega]

13 minutes ago, maddangerous said:

 

Very true, just don't have a lot of time at the moment (we just had a baby recently). 

I feel like the new cpu message isn't a good sign though, honestly. the only time i've seen that in my experience, was failing hardware. Certainly doesn't mean that's always the case though.

 

I'm going to at least pull RAM, and shut off all of the M.2 drives except boot (all the drives are under M.2 shields built on the motherboard, but I can disable them in the EFI). I don't have any SATA devices.

 

Board has a dual-bios, but I have to pull some things to get to the switch.

 

 

So, while I haven't pulled the CMOS battery, there is a reset to defaults type of button on the I/I shield, I just tried that a few minutes ago to no avail.

 

 

It's just so weird, I've had 0 problems this entire time. Built the rig on February-ish, and I never enabled AMP due to the BIOS bugs. when AGESA 1.0.0.7c came out, after the fix, I enabled AMP. Checked voltages in hwmon64, everything looked fine.

You should try removing the battery despite the button on the back. 

[pio]

Remove battery.
Reseat the everything.
Once its working again, maybe try to update the BIOS if possible?

Idk, stuff like that I typically just "throw the book at it" so to speak, and just try everything till it works right.  I doubt its dead.

[maddangerous]

1 hour ago, UltraMega said:

You should try removing the battery despite the button on the back. 

 

I will do this when I can get to it, for sure.

 

1 hour ago, pioneerisloud said:

Remove battery.
Reseat the everything.
Once its working again, maybe try to update the BIOS if possible?

Idk, stuff like that I typically just "throw the book at it" so to speak, and just try everything till it works right.  I doubt its dead.

 

I do want to try to update the bios, I have to be stable enough to download it though.

[bonami2]

May be a bent pin reseat the CPU and look at the pin. 

[Sir Beregond]

If I am reading this right, are you saying it started after enabling AMP following an AGESA update?

[maddangerous]

11 hours ago, bonami2 said:

May be a bent pin reseat the CPU and look at the pin. 

 

Just realized this wasn't in OP, but it's been running fine since about february-ish when I built it. That said, I would think a bent pin would present issues sooner? Never have had a pin bend so, just guessing.

 

1 hour ago, Sir Beregond said:

If I am reading this right, are you saying it started after enabling AMP following an AGESA update?

 

Close, Updated AGESA and BIOS in the last week of august (waited about a week and a half after the release of that BIOS version, which was August 16th in my case). Enabled AMP, and everything had been fine. Voltages were in spec.

 

Memory Context Restore has been enabled... I believe as long as I've had the build.

 

I enabled MSI's high efficiency mode and set it to the "tighter" setting, I believe on Saturday.

Edited October 26, 2023 by maddangerous

[ENTERPRISE]

Whilst this will be an unlikely "magic" fix. I would suggest using the BIOS flashback feature with the latest BIOS downloaded from another computer. The 1.0.0.8 AGESA BIOS should be released for your mobo by now I would imagine.

[maddangerous]

24 minutes ago, ENTERPRISE said:

Whilst this will be an unlikely "magic" fix. I would suggest using the BIOS flashback feature with the latest BIOS downloaded from another computer. The 1.0.0.8 AGESA BIOS should be released for your mobo by now I would imagine.

 

That's in the cards to try, and yeah it was released oct 17th as a beta bios. Just have to get the time to get to it.

[maddangerous]

Aight so, I had to reset the BIOS twice, but I was able to get into windows.

 

First reset, post code checker was stuck on 00 for a few minutes, and one of the debug LEDs was on (CPU I think?)

 

Power cycled the machine, same thing, so I did a BIOS reset again and I was able to POST + load windows.

 

First thing I did, was make a backup of my stuff. It wasn't time for my weekly backup yet, so I had some stuff I'd lose if I didn't do that, so I'm good there.

 

I'm going to grab the latest BIOS and flash that, see what happens.

 

Oddly, the whole machine does feel snappier overall, so something funky was definitely happening. I didn't change any BIOS settings, prioritized getting my data before messing around.

Edited October 27, 2023 by maddangerous

[UltraMega]

1 hour ago, maddangerous said:

Aight so, I had to reset the BIOS twice, but I was able to get into windows.

 

First reset, post code checker was stuck on 00 for a few minutes, and one of the debug LEDs was on (CPU I think?)

 

Power cycled the machine, same thing, so I did a BIOS reset again and I was able to POST + load windows.

 

First thing I did, was make a backup of my stuff. It wasn't time for my weekly backup yet, so I had some stuff I'd lose if I didn't do that, so I'm good there.

 

I'm going to grab the latest BIOS and flash that, see what happens.

 

Oddly, the whole machine does feel snappier overall, so something funky was definitely happening. I didn't change any BIOS settings, prioritized getting my data before messing around.

 

Now that your back into windows, after you update your bios you should run some stability tests. I like to run prime95 or CPUz stress and furmark at the same time. 

[maddangerous]

3 minutes ago, UltraMega said:

 

Now that your back into windows, after you update your bios you should run some stability tests. I like to run prime95 or CPUz stress and furmark at the same time. 

agreed. However, why furmark? Isn’t that GPU?

[pio]

42 minutes ago, maddangerous said:

agreed. However, why furmark? Isn’t that GPU?

Full system test is what @UltraMega was getting at I believe; p95, OCCT, Cinebench, or CPUz (CPU stability test) added onto it would be a full system test where you have everything loaded at 100%.  So, yes Furmark would be the GPU portion.  A full system test would be a test to make sure your PSU is working appropriately.  OCCT also has this function as a built in test too if you'd rather do it all in one place.

[maddangerous]

well.. I didn't get the change to run a full system stress test.

reinstalled windows since that was corrupted, and it froze. Upon reboot, intermittantly, I'll get the CPU and MEM debug LED's on the board turning on and it fails to boot.

 

If it doesn't fail to boot, windows bluescreens, and back to the debug LEDs again.

 

I don't have another AM5 CPU to test with, and moving DIMMs around didn't help.

 

Looks like I'm going to Microcenter tomorrow, hopefully whatever the hell happened didn't damage all 5 of my NVMe drives too.

Edited October 30, 2023 by maddangerous

[pio]

30 minutes ago, maddangerous said:

well.. I didn't get the change to run a full system stress test.

reinstalled windows since that was corrupted, and it froze. Upon reboot, intermittantly, I'll get the CPU and MEM debug LED's on the board turning on and it fails to boot.

 

If it doesn't fail to boot, windows bluescreens, and back to the debug LEDs again.

 

I don't have another AM5 CPU to test with, and moving DIMMs around didn't help.

 

Looks like I'm going to Microcenter tomorrow, hopefully whatever the hell happened didn't damage all 5 of my NVMe drives too.

I don't know if this helps you out at ALL with your problems or not, but I had something similar happen.

I had a B450 board, Ryzen 2600 (kid's rig).  I was needing a CPU to test a new B350 board I got in, so I pulled the 2600 out to test with.  When I was done testing the new B350 board, I accidentally dropped my Ryzen 2600 onto the (new) B350 board, and it bent a few pins on the CPU.  I did attempt to straighten them out.  Then, I tried to put the CPU back into my B450 board.  Not knowing there were still bent pins, I was never fully able to get the CPU seated properly again in the board.  HOWEVER, I still tried to get it to work anyway, because I could've SWORN I got the pins straight and the CPU was in there right.  Kept trying, and trying.  Eventually, gave up and bought him a new CPU.  Popped the new 2600x into the B450 board, and it worked..........kinda.  Random freezing, couldn't figure it out.  Bought a new motherboard, put everything into that, and his rig works great ever since.  Now.........I bought a Ryzen 1200 recently.  I tried to pop it into that same old B450 board since I had it laying around.  Yeah.........same problem, random freezing.........AGAIN.  

It very well could be your motherboard, and you might not have actually done anything that could've damaged it that you are aware of.  I've NEVER heard of ANYONE hurting an AMD socket before, never.  I did it.  I've also never heard of anyone dremeling through a GPU PCB before, or burning a Gigabyte motherboard to the ground in literal flames, bending the PCB on a 6900XT with thermal pads..............but I've done it.......... 

Edited October 30, 2023 by pioneerisloud

[maddangerous]

1 hour ago, pioneerisloud said:

I don't know if this helps you out at ALL with your problems or not, but I had something similar happen.

I had a B450 board, Ryzen 2600 (kid's rig).  I was needing a CPU to test a new B350 board I got in, so I pulled the 2600 out to test with.  When I was done testing the new B350 board, I accidentally dropped my Ryzen 2600 onto the (new) B350 board, and it bent a few pins on the CPU.  I did attempt to straighten them out.  Then, I tried to put the CPU back into my B450 board.  Not knowing there were still bent pins, I was never fully able to get the CPU seated properly again in the board.  HOWEVER, I still tried to get it to work anyway, because I could've SWORN I got the pins straight and the CPU was in there right.  Kept trying, and trying.  Eventually, gave up and bought him a new CPU.  Popped the new 2600x into the B450 board, and it worked..........kinda.  Random freezing, couldn't figure it out.  Bought a new motherboard, put everything into that, and his rig works great ever since.  Now.........I bought a Ryzen 1200 recently.  I tried to pop it into that same old B450 board since I had it laying around.  Yeah.........same problem, random freezing.........AGAIN.  

It very well could be your motherboard, and you might not have actually done anything that could've damaged it that you are aware of.  I've NEVER heard of ANYONE hurting an AMD socket before, never.  I did it.  I've also never heard of anyone dremeling through a GPU PCB before, or burning a Gigabyte motherboard to the ground in literal flames, bending the PCB on a 6900XT with thermal pads..............but I've done it.......... 

 

Wow, that sounds like an adventure lol

 

Honestly, i have a warranty, and if I'm getting it to the point where I'm checking the socket... I'm RMA'ing it.

It's been rock-solid until wednesday, so I'm doubting bent pins. At this point, I'm honestly unsure what happened.

[J7SC_Orion]

6 minutes ago, maddangerous said:

 

Wow, that sounds like an adventure lol

 

Honestly, i have a warranty, and if I'm getting it to the point where I'm checking the socket... I'm RMA'ing it.

It's been rock-solid until wednesday, so I'm doubting bent pins. At this point, I'm honestly unsure what happened.

 

I can certainly understand your frustration... Before RMAing, I would do one more attempt to completely reset everything (ie. battery out, PSU disconnected) because you might be stuck in no-man's land on the bios chip itself. Presumably, the board also has bios flashback - going back to the earlier bios version, if possible. If that doesn't solve it, you probably do want to RMA, but is it the CPU, the mobo or RAM ?

[maddangerous]

Alright I've been kinda bad at keeping this updated.

 

I've got the BIOS reset, and set these settings:

 

 

I'm not running the speed and timings this kit is capable of at the moment.

 

After settings were changed I was able to boot.

 

I reinstalled Windows 11 Pro, ran all windows updates and installed all of my motherboard drivers (Several reboots during this time.)
I went to install the Corsair iCue software, as I have an HX1000i PSU and Virtuoso XT RGB.

 

After that installed, I almost immediately got a BSOD for MEMORY_MANAGEMENT, and when I rebooted from that, another BSOD IRQL_NOT_LESS_OR_EQUAL, and the system fails to get past initial Windows boot stages. I got one more MEMORY_MANAGEMENT BSOD, and another BSOD IRQL_NOT_LESS_OR_EQUAL.

 

All caused by ntoskrnl.exe, none have anything corsair indicated in the bugcheck files. Even the same memory address for them. Keep in mind that this is a ~20 minute old Windows install, from a known good USB stick that I made on a separate working computer.

 

When I booted into the BIOS, and reset to factory default settings, rebooted back into the BIOS immediately and set those same settings as in my screenshot above, computer seems to be fine since, the little bit I've used it.

 

This is very strange. I'll be running stress tests before moving on to buildzoids timings and getting this kit off JEDEC speeds.
I haven't run any stress tests yet, that's on the list for this weekend.

 

 

Edited November 3, 2023 by maddangerous

[pio]

I don't know if this helps at all, but if you have any SATA drives, have you tried swapping SATA cables?  You've got me stumped, that's really weird that it was fine till IceQ was installed too though.  

[maddangerous]

3 hours ago, pioneerisloud said:

I don't know if this helps at all, but if you have any SATA drives, have you tried swapping SATA cables?  You've got me stumped, that's really weird that it was fine till IceQ was installed too though.  

 

Nope, no SATA drives. All NVME.

I too am stumped. although, I'm not all the way through testing, so there's the possibility I'm on the right track here.

 

going to put this through some stress testing later today and see where that gets me. hoping it's stable, fingers crossed. I'd like to get this kit running the speed/timings it should be running

 

[UltraMega]

Sounds like a bad mobo to me.