9800X3D -- Entire system freeze

[Kirusei]

Built a new PC at the start of the year and have been dealing with this one issue since February that has been driving me insane and have now been throwing money at the situation hoping it would get fixed.

Ever since the second open beta for Monster Hunter Wilds, I'd have a big issue where my entire system would freeze after a while of gameplay and I'd have to manually restart my PC. Unfortunately since it froze, no logs in Event Viewer were made to record what happened, and Reliability Monitor only came up with "Windows didn't shut down properly" with no extra information. I first thought it was my undervolt settings causing it since the issue looked similar to what I ran into when trial and error'ing my PBO settings, but I kept running into the same problem even after disabling any overclock settings. Eventually I gave up and waited for the game to release instead, hoping that it was just beta shenanigans

Once the game released, I'd run into issues similar to what I experienced in the beta test. I got frustrated after the issue occurred a few more times and decided to take time off that week to narrow down the problem/s. The first few things I did was clear the CMOS and a fresh Windows install, which unfortunately didn't fix the freezing. After that, I thought that I may have defective hardware and issued an RMA for my CPU and made plans to get a replacement motherboard and RAM. While I waited for the new parts, I tested my GPU and storage drives by putting them back in my previous PC build and confirmed that they weren't the cause.

Once all the replacement parts came in I tested them again, only to run into yet another system freeze while benchmarking Wilds. Decided to post on a forum about this ordeal, and someone mentioned that my CPU is recommended to be water cooled, so I ended up buying an AIO cooler.

Yesterday, I just installed the AIO cooler only to eventually get another system freeze after hours of testing with the benchmark, and the more I test it, the less time it takes for the game to freeze my system. I'm now down to my last straw thinking that I might need to buy a new PSU. I've been debating with myself on either if something within Wilds is causing this unintentionally, or if my hardware may be faulty especially after a fresh install of Windows which usually fixes my software problems.

Is anyone else experiencing this issues, and has a fix been found? If you have experienced it, drop your CPU, motherboard model, RAM, and CPU cooler in the replies. I wanna try to see if there's a sort of trend

9800X3D
MSI MAG B650 TOMAHAWK
DDR5 G.Skill Trident Z5 Neo RGB 32GB(2x16) 6000MHz CL30
Noctua NH-D15S/ Be Quiet! Silent Loop 3
Straight Power 11 750W Platinum
4070 Ti Super

[Kirusei]

Originally posted by アンジェル:

Originally posted by Kirusei:

I've ran stress tests like OCCT and Y-Cruncher for both my RAM and CPU and didn't get any errors.

So no dedicated RAM test.

Was able to finally get memtest to work earlier this morning, both RAM sticks passed

[Kirusei]

Originally posted by defnotj4:

probably more worthwhile asking OP if the power supply is actually new as well with this build or a carry over from something they made pre covid?

it doesn't look like this is even sold anymore (just glancing at amazon) and and an aging PSU not reliably pumping enough juice sounds about right here

Yeah, this was carried from my previous PC which ran a 5800X. Never had any problems with it in my previous build. While I don't think it matters much, the motherboard in my new build has a 16-pin CPU connection and I can only take up 8 of those with this PSU.

Originally posted by Turbo Enjoyer:

Did you try a bios update? seems like that b650 boards does not fully support 9800x3D out of the box.

I've checked multiple times to see if there's any new updates, but yes my motherboard has an updated BIOS

[アンジェル]

Originally posted by Kirusei:

Originally posted by アンジェル:

So no dedicated RAM test.

Was able to finally get memtest to work earlier this morning, both RAM sticks passed

That is good news! So next would be to check the theory about the power, I guess.

[Zatick]

Your PSU should have three 12V P8 connectors spread across two cables. If you were going to connect both sockets on the motherboard you'd want to use both cables. You "probably" don't need to populate both sockets unless you're running a fairly heavy overclock, which I understand you're not doing here.

But if you scroll back a bit, if you are not already, it's absolutely necessary to power the GPU from both PCIe 6+2 cables. So plug both cables into your PSU and then use one 6+2 plug on each cable to plug into the GPU adaptor.

https://imgur.com/a/5hIZuDW

[Kirusei]

Originally posted by Dae:

Not defending this game but if you're using PBO all core -30 with x10 scalar like all those silly guides suggest, tone either the negative offset down or reduce scalar or both, or better yet do it properly via per core. Make sure your SOC isn't too low or high also. Anything under 1.25v is safe, though you shouldn't need as much as that anyway (for reference I'm at 1.175v 6000mhz CL30 too but guarantee it'll be different for you).

Had the same freezes in Elden Ring when trying to find the sweet spot despite synthetics claiming I was stable. I was able to dodge the hard reset by simply hitting windows key and opening a Google Chrome tab which weirdly allowed the system to recover.

Nah, I've understood that those people only use Cinebench as their benchmark which doesn't really help finding a stable curve if it's only one benchmark you're using. Also I haven't used PBO ever since running into my issue for the first time just to minimize and other problems

[Kirusei]

Originally posted by Zatick:

Your PSU should have three 12V P8 connectors spread across two cables. If you were going to connect both sockets on the motherboard you'd want to use both cables. You "probably" don't need to populate both sockets unless you're running a fairly heavy overclock, which I understand you're not doing here.

But if you scroll back a bit, if you are not already, it's absolutely necessary to power the GPU from both PCIe 6+2 cables. So plug both cables into your PSU and then use one 6+2 plug on each cable to plug into the GPU adaptor.

https://imgur.com/a/5hIZuDW

I'll check it out soon

[Kirusei]

Originally posted by Kirusei:

Originally posted by Zatick:

Your PSU should have three 12V P8 connectors spread across two cables. If you were going to connect both sockets on the motherboard you'd want to use both cables. You "probably" don't need to populate both sockets unless you're running a fairly heavy overclock, which I understand you're not doing here.

But if you scroll back a bit, if you are not already, it's absolutely necessary to power the GPU from both PCIe 6+2 cables. So plug both cables into your PSU and then use one 6+2 plug on each cable to plug into the GPU adaptor.

https://imgur.com/a/5hIZuDW

I'll check it out soon

Just connected both VGA1 and 2 through the adapter and am gonna start the benchmark to see if the issue still persists

wish me luck

[Kirusei]

Originally posted by Kirusei:

Originally posted by Kirusei:

I'll check it out soon

Just connected both VGA1 and 2 through the adapter and am gonna start the benchmark to see if the issue still persists

wish me luck

Ran into a system freeze after an hour of testing unfortunately

[Zatick]

Well given the nature of the crash it does sound like it's a power issue, but not necessarily PSU issue.

What motherboard BIOS version are you running?

[Kirusei]

Originally posted by Zatick:

Well given the nature of the crash it does sound like it's a power issue, but not necessarily PSU issue.

What motherboard BIOS version are you running?

1.M3, the most recent

[Zatick]

Hmm, well try to disable AMD EXPO on the RAM and see if that helps. I've seen several systems where EXPO or XMP (Intel equivalent) had these sorts of issues.

Apart from that, if all of the drivers (in particular chipset driver) are up to date, then it might be time to start looking at replacing the PSU.

[Turbo Enjoyer]

If the PSU change still doesn't solve it, try a different motherboard. I have the same RAM and CPU but my board is an x870E

I also don't think EXPO and PBO are the issue. I had both enabled and can play the game without your issue. PBO is just set to -30 curve optimizer to reduce some heat since I'm just using an air cooler.

[Zatick]

PBO with an aggressive undervolt has been known to cause instability, and it's not consistent across all processors of the same model. Same as overclocking and some individual CPU's needing more voltage than others.

The RAM, it really depends on the manufacturer and how well their QC is doing at the time. All of this RAM is just binned, you see all of the models with different timing and clocks, they are the same memory modules. They test the individual units and determine their max performance and then sell them based on their capability, the same with "Super" GPU's that offer marginally higher clock rates etc. I have had more than one set of G-Skill RAM that didn't work at the stated XMP speed/timing, had to drop it down to SPD speed/timing, but that was a fair while ago though.

It's really just a process of elimination, if the problem goes away as soon as you relax speed/timings, or give something a bit more voltage, then you know where the problem is. The thing is he's replaced the motherboard, so it doesn't seem likely it would be motherboard hardware failure. The RAM seems to be on the compatibility list for the board, and the G-Skill compatibility checker also reports it's compatible with that specific board.

[Kirusei]

Looks like I'm getting a new PSU

[Jeff4u]

Originally posted by Kirusei:

Looks like I'm getting a new PSU

Better cap the fps in game benchmark to see its stability to see it is actually PSU. Anyway, use game to test may not be good idea.

If it passed, run CPU, GPU and memory stress stability test at the same time to check whether same PSU problem.