5800x stability issues

[Aretak]

There's no LLC option in this mobo's bios, i'd have to set voltages somewhere else. Don't mind tweaking voltages or curve optimiser settings to extract performance but having to tweak to get stability suggests a faulty CPU or some other hardware issue

There is. At the top of the main OC page, flip the setting from 'Normal' to 'Expert' and it'll expose the 'DigitALL Power' menu where you can set it.

 

[BUDFORCE]

I think you might want try reinstalling Windows, sorry I know you said it's pain but odd how you can bench without fail but you are getting issues in games, maybe software/driver failure somewhere.

Infact, have to tried clean graphics drivers?

 

[philo-sofa]

to answer a few of these reponses (thanks for taking the time guys)

Wouldn't normally use a beta bios but this one seems to offer some stability albeit at the cost of some performance

SFC /scannow didn't find any errors

Already using HWInfo, just found it odd that the beta bios prevents RyzenMaster reading some of the chip's telemetry, clearly a bug in the bios

There's no LLC option in this mobo's bios, i'd have to set voltages somewhere else. Don't mind tweaking voltages or curve optimiser settings to extract performance but having to tweak to get stability suggests a faulty CPU or some other hardware issue

Regarding the PSU, replacing this would mean chucking money at the problem. I know there's a chance it's the weak link but it's never given any issues before and it's well within spec for a 3080 + 5800x. I really really don't want to get into a loop of replacing other parts in the hope they fix this. Is there anything that can log the voltages coming out of the PSU in heavy-load situations to establish it as a possible culprit?

Hey read my post: enable PBO with a positive curve:

Okay so you've run the memory at JDEC too....

One more idea: the crashing you've experienced does sound a bit like it's failing when returning to a more idle state which is a common issue for Zen 3. Try PBO set to manual and set an all core curve of +15. Out of the box the internal curve the chip is using may be too aggressively negative for the silicon.

 

[opethdisciple]

You mention VR, does this BSOD happen whilst you are playing a VR game or does it also happen during more normal 2d gaming sessions?

Could be a dodgy peripheral or peripheral driver.

 

[varkanoid]

Regarding the PSU, replacing this would mean chucking money at the problem. I know there's a chance it's the weak link but it's never given any issues before and it's well within spec for a 3080 + 5800x. I really really don't want to get into a loop of replacing other parts in the hope they fix this. Is there anything that can log the voltages coming out of the PSU in heavy-load situations to establish it as a possible culprit?

I have a PSU tester I bought probably nearly 15-20 years ago now it cost me a tenner from the blue shirt guys when they used to have stores and I have probably used it about 2 or 3 times (plus in a custom water loop to test it before it went in my PC instead of the paperclip method) and even though I've used it so little its probably saved me about £200-£300 in PSU costs as it ruled out an issue with the PSU.

Pretty sure you can still get them cheap in one guise or another.

 

[jim_britton]

Had very little time to pull the rig apart and test stuff this weekend. I still intend to roll my sleeves up, starting with the CPU voltage settings in BIOS (LLC and offsets).

In the mean time I reverted to the latest beta BIOS with the lower boosting frequencies so I could do my league racing (yeah yeah, I'm a massive nerd) and had no issues. Interestingly, comparing the real-world performance (VR frame times) with the high-boosting-unstable bios with the same for the low-boosting-beta-bios, there's just no measurable difference at all.

At this point I'm seriously tempted to chalk it up to a PSU that's not as good as it ought to be, with a viable workaround of the low-boosting beta bios. Maybe when life is a bit less frantic I can systematically test the various voltage settings and establish if there's any that give stability with the non-beta BIOS, or maybe treat myself to a PSU. I also need to reseat the chip, reinstall windows, try single RAM stick, and a million other things that I just don't have time for

 

[jim_britton]

Took a little time to investigate this further, following the various advice on here and some other bits and bobs I've found (like setting the idle power to 'typical current' or some such).

The end result is that even with a posh new PSU (Corsair RMx 850), clean Windows install, single RAM stick, default settings, various BIOS versions, reseating the CPU and spending hours and hours of time that I would rather be doing something fun, I *still* can't get this 5800x to be stable. It's a real sod to test because it only ever crashes when gaming so no synthetic benchmarks (Prime95, Cinebench, OCCT, CPUZ) can reproduce the crashes.

The only think I've not tried is swapping my motherboard, but having thrown away £130 replacing a perfectly adequate PSU an spending a couple of days on this already I really don't want to do this. When I put the 3700x back in it was stable

 

[VonScar]

RMA that CPU. The 5800X.

 

[Dg834man]

Rma it had the same thing with a 3800x

 

[SilverCider]

Have you tried GPU benchmarking, or a mixed bench that'll cover both CPU and GPU. The 5800X will be enabling the 3080 to work harder more of the time during gaming and so it could be the culprit. It seems odd that you can peg your CPU and/or memory at 100% through the various CPU benchmarks and it will be fine but mixed scenarios such as gaming will cause your BSOD. Something else to ponder before RMA'ing your CPU.

 

[Moeks]

It's a bit academic now you've replaced the PSU, but I was going to say, before you draw too many conclusions about it's true capabilities, check how you have it actually hooked up to the 3080, are you using a single PCI-E power cable and using the daisy chain to connect both connectors on the graphics card, or are you running each PCI-E power cable separately back to the psu, I would strongly suggest the later I have read a number of accounts of people doing the former and having problems as it will pull a lot of juice. Assuming it was running in exactly the same configuration with the 3700X though and you didn't put the 3080 and the 5800X in together then I doubt it's the problem, it sounds like it might well be the CPU from what you have said.

 

[LadFromWales]

Is the GPU overclocked? If it is trying running it at stock settings.

If it is still crashing I'd looking at the CPU at some point in its boost curve is requesting a voltage too low for a given frequency, which in your case isn't a frequency utilised while idling or while maxing it with P95. You could verify this by setting a positive cpu core voltage of perhaps 0.05V, or positive values in core optimiser as others have said.

If you're having to do this though I'd be looking at either your board being naff and not supplying enough voltage when the CPU wants it, or the CPU being faulty. What is the board? Do you have another board or a friend's system to test with?

 

[jim_britton]

I've continued to monitor and faff with this and am still undecided about it.

At the weekend I had some regular crashes in Dirt Rally 2 and decided it could only be the CPU and contacted OC to start the RMA process. I got prompt a reply back from OC asking for a bit more info and this made me want to do just a tiny bit more testing, with some interesting (to me at least) results.

At stock settings - XMP off, most stuff on 'auto', processor virtualisation on (I need this cpu for VMs) - it crashes with WHEA bugcheck error 0x0000124.
If I enable PBO and set the PBO limits to Motherboard, with no other changes, it *doesn't* crash
If I enable XMP with PBO limits set to Motherboard it crashes with WHEA Bus/Interconnect Error (on various APIC IDs)
If I leave XMP enabled and set the IF frequency down to 3200 (from 3600, which is what it's set to when I enable XMP) it *doesn't* crash

Note that in each case the crashes only ever happen when gaming. Crashing happens within about 15 mins, the doesn't-crash cases are stable for at least a couple of hours.

So it appears there are 2 triggers for these crashes - the CPU isn't stable with IF set to 3600, and the CPU isn't stable with PBO limits set to auto. With XMP on, IF at 3200 and PBO limits to Motherboard (which means crazy PPT limit like 500w, EDC at like 200 or something) the thing appears to be stable in gaming.


To answer some of the questions here, the GPU is hooked up to 2 separate PCIe power connectors (as it was before), no daisychaining. It's not overclocked. The board is an MSI x570 pro-a. I know this has relatively weak power delivery gubbins but would still have expected it to be able to run a supported CPU at stock settings.

I don't have access to another board to test. I've not tried positive voltage offset

 

[alec]

"XMP on" crashing at 3600 is probably not faulty CPU, just settings. Should be fixable by tweaking voltages or timings, not necessarily dropping IF.

The "XMP off everything auto" crash is the most suspicious. And if it is really fixed by enabling PBO, its the motherboard doing something wrong on Auto, not CPU.

 

[jim_britton]

I do agree, it's starting to smell a bit like the motherboard

 

[LadFromWales]

OK, try IF 1800 1:1, and set memory timings to 16 20(both tRCDs) 20 40and tCWL to 16, memory voltage to 1.4v.

Reset CMOS before hand, change nothing else, except virtualisation I guess if you need that while testing.

Even 18 22 22 42 tCWL 18 if you want to be sure its not the RAM, but I'd try the above first.

This should rule out the board setting timings that it and the IMC/RAM aren't happy about.

I've known boards to fall over at XMP, yet set the exact same settings as it sets manually, all is well. But at least the above will ensure you are able to run 1800 in sync with the IMC.

 

[Dave2150]

Sorry to hear of your troubles. For what it's worth, one of my mates had a similar issue with a 5800X system. He spent weeks troubleshooting, eventually RMA'ing the board and CPU, only to find his issues were due to an incompatibility with USB devices he was using.

He ended up selling the whole system and bought a Intel 11400 CPU system instead. He made money on the trade, though more importantly, hasn't had one BSOD or crash of any kind. Just works, he doesn't overlock, so just had to set XMP and enjoy the 24/7 stability.

 

[jim_britton]

Arrggghhhh.... wasted too many evenings on this. Turns out it *does* still crash with PBO on. Tried manual RAM timings, no change. Re-seated and re-pasted again (not in that order, obviously), ran for a while with XMP disabled and IF at 1333 - still crashed (but less). Re-enabled XMP, ran IF at 1800 (ram at 3600) but with Global C-States disabled and *thought* it was stable, but no. 2 crashes in quick(ish) succession running Dirt Rally.

And probably a hundred other things I've tried this week. Put the 3700x back in and all is peachy. Except for this £400 paper weight. I give up. Hopefully Overclockers will accept all this endless faff-and-fail is sufficient evidence that the chip is clearly a basket case and replace it

 

[Grim5]

Arrggghhhh.... wasted too many evenings on this. Turns out it *does* still crash with PBO on. Tried manual RAM timings, no change. Re-seated and re-pasted again (not in that order, obviously), ran for a while with XMP disabled and IF at 1333 - still crashed (but less). Re-enabled XMP, ran IF at 1800 (ram at 3600) but with Global C-States disabled and *thought* it was stable, but no. 2 crashes in quick(ish) succession running Dirt Rally.

And probably a hundred other things I've tried this week. Put the 3700x back in and all is peachy. Except for this £400 paper weight. I give up. Hopefully Overclockers will accept all this endless faff-and-fail is sufficient evidence that the chip is clearly a basket case and replace it

Time to rma, overclockers can test in their own system for stability

 

[StephenJ]

looks like you have done a good amount of testing to show its the cpu time to RMA