Question Did I just brick two 7900xs?

[RogerDorsey]

I recently built a new PC, and immediately started struggling with freezes. After switching out various components, I came to the conclusion that my CPU wasn't working correctly, and so I ordered another one thinking I'd return the old one later. (I'd already tried different RAM, motherboard, GPU and PSU)

Unfortunately, I've just put the new one in, and my system froze almost immediately. I rebooted and ran memtest86 and received plenty of errors. I tried putting my old chip back in, and now it's behaving much worse than it was before. I can't even make it to my OS 9/10 times. I've put the newer chip back in again as well, and now that won't even boot up either.

To give some background, when I first built my PC I was getting a full on system freeze every few hours. I was also getting a lot of programs crash with memory access violation errors. If I recall correctly, at some point I re-seated my CPU, and things improved somewhat. The last week or so I didn't have a single permanent freeze, and I only had 1-2 instances of a program crashing.memtester was still flagging errors when I ran a lot of tests, but it was rare.

The fact that this same CPU is now causing a crash every minute or so is worrying me a bit. I'm fairly confident it's not an overheating issue, so given that the chance of buying two faulty CPUs is probably rather high, I can only assume I've done something wrong.

The only things I can think of is that I've maybe applied too much thermal paste. There is a bit that has come down over the sides of the heat sink element and one or two small accidental smears on the edges of the chip itself (see here). Either that, or I've somehow killed them both with static, which seems quite unlikely? I'm no expert, but I've built around 10 machines before and never had any problems.

Does anyone have any suggestions on what I could possibly try next? Presumably I'm not going to be able to get a refund on two 'faulty' CPUs from the same retailer, so I'm feeling pretty silly right now. In case anyone is wondering, I didn't attempt any overclocking or anything like that. I did not mess with any settings in the BIOS at all.

My specs:

• AMD Ryzen 9 7900X
• Gigabyte MB GBT AMD AM5 B650M DS3H
• 2x16GB DD5 memory

 

[Ralston18]

Update your to include full system hardware specs and OS information.

PSU(s): make, model, wattage, age, condition (original to build, new, refurbished, used)? History of heavy gaming use, video editing, or even bit-mining?

Did you use PSU cables from any other PSU other than the specific installed PSU?

Disk drive(s): make, model, capacity, how full?

Where were the 7900's purchased?

Thermal Paste: source, brand, how applied?

RAM: likewise, brand and source?

 

[RogerDorsey]

I have two SSDs, one has Linux Mint, the other Windows 10. I can't boot into either OS at the moment.

This is the PSU I'm using (650 Watt version) I'm only using the cables that came with it. I've also tried my old PSU to see if that helped, which was a 550Watt OCZ ZS, but no joy.

My main 2TB SSD, which runs Linux Mint, is a few weeks old and has plenty of space. It's one of these

Both CPUs were purchased from Amazon, described as being new. One from a third party seller, one directly from Amazon EU from what I can tell.

The first CPU I covered in the paste that came with my cooler (Cooler Master Hyper H412R CPU Air Cooler). The second one I used some Arctic MX-4 paste. I basically put a generous blob in the middle of the CPU, then spread it around a little bit with the end of the syringe before putting on the cooler and screwing it in.

I've tried two different sets of RAM, both 2x16GB. Exact models are here and the other set was 'Kingston FURY Beast DDR5 32GB (2x16GB) 5200MT/s DDR5 CL36 DIMM'. Again both sets were new from Amazon. I've tried individual sticks as well just to make sure. I'm fairly confident the RAM must be OK.

I'm also running a Nvidia 1070, but I've also tried just removing it altogether and running with the integrated graphics chip. Still had problems

 

[pete_101]

With my recent experience of Gigabyte B650 boards, my first suggestion is try a different BIOS version. On a B650m gaming x I tried all 6 BIOS versions which had been released for it, only 1 worked without some kind of issue.

 

[RogerDorsey]

I've done some further testing to try and rule some things out.

I ran just a motherboard, CPU and 1 stick of RAM. No HDD, no extras, just a USB stick with memtest86. The 1st of 4 runs passed, but then I got my first error after 55 minutes. That seems a lot better than earlier when I was running into multiple errors within minutes.

I'm now running on 1 stick of RAM with my SSD and GPU, and I've managed to at least boot into Linux. I'm still getting errors when running memtester, but for now at least my PC remains up.

At the moment I've got one 7900x in a Gigabyte B650, with the Patriot Viper RAM.
The other 7900x is in an MSI PRO B650M-A, with Kingston Fury RAM.

Both setups appear to have the same problem.

It could be dropping to 1 stick of RAM, not plugging in the case USB connectors, the HD audio pins, or plugging in fewer hard drives, but both setups have at least allowed me to boot into my OS and go error free for at least 10 minutes. I suppose that tells me the CPUs aren't completely bricked!

 

[RogerDorsey]

With my recent experience of Gigabyte B650 boards, my first suggestion is try a different BIOS version. On a B650m gaming x I tried all 6 BIOS versions which had been released for it, only 1 worked without some kind of issue.

That's interesting. I've been trying out two different motherboards, but despite one being Gigabyte and the other an MSI, they both are B650m...

 

[RogerDorsey]

So last night I rebooted my machine and upped it back to 32GB of RAM. Since then it's been running without any problems whatsoever. I'm running memtester over and over and not finding anything.

I really can't explain it. Yesterday I spent hours trying both CPUs and with every attempt I was getting freezes after a minute or so. It feels like every time I tinker with my build slightly I get a random chance of having problems or not.

The fact I was getting memtest86 errors with a bare-bones setup is leaving my baffled still.

 

[RogerDorsey]

Just in case anyone comes across this in the future. I think I solved the problem.

In my BIOS there was some sort of turbo boost setting that claimed to overclock the CPU when it sensed a good opportunity (e.g. low temperatures)

I changed this setting from 'default' to 'disabled' and haven't had a problem since.

 

[Avro Arrow]

Yeah, I was going to say that it sounds like a motherboard problem based either in the BIOS or the power distribution. Looks like you got it handled though!