Processors

[Louier]

I've tried for the past week figuring out the problem with my computer, seemingly randomly it restarts in games, on browser etc.

My Specs

- AMD 5900X

-MSI x570 Unify

-Nvidia 3080ti

-32gb corsair dominance ram

-850w BeQuiet Straight Power PSU

-3tb WD hdd, 500gb samsung evo 870

Windows 10

Errors - Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 8

-Error setting traits on Provider {8444a4fb-d8d3-4f38-84f8-89960a1ef12f}. Error: 0xC0000001

[goodplay]

Try checking that you have optimal bios settings, either msi forum or overclockers.uk/overclock.net 5900X owners threads.

Is your ram on the mobo qvl and not mixed sets.

And have a read of hwinfo forum power deviation thread.

[kurdtnz]

Is it a new build or one you've had for awhile that has only just started playing up? If it is a new build, did you update the BIOS? (some B550 and X570 boards still need a BIOS update to run 5000 series cpu's.)

[Louier]

No it’s not a new build I’ve had the cpu for awhile, I flashed my bios when I installed the cpu, recently I upgraded my card to a 3080ti, it’s been working as normal for around a month then suddenly started restarting recently. I also just got another error - 

A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 4

https://imgur.com/a/gWWdjFX 

[linkz0rz]

Could be your PSU, change it for a different more powerful PSU from a reputable manufacturer (not Gigabyte) like Corsair, it might fix the problem

[Louier]

I was thinking about this but it should work with 850w, I was hoping to see if there could be another issue before changing psu

[linkz0rz]

I had a 1000w ThermalTake PSU I thought was good enough for my RTX 3090 but my pc kept restarting and going to black screens every time I tried to load up any power intensive game like Cyberpunk 2077. I also didn’t believe it was my PSU until I changed it for a Corsair RM850x 850W Gold PSU and my pc was stable af. 

[Louier]

Yeah I'll probably get a new psu, just wondering if it could be a heat related problem with the gpu.

[Cmdr-ZiN]

@Louier There many threads on this issue, just look for WHEA logger errors, Ryzen instability and error id 6008 etc.

There's many people affected by Ryzen instability it's been around since the 3000 series but seems worse witht he 5000 series.

I'm on my 3rd CPU, my 2nd CPU was actually perfect for a couple of months and then it completely died, it never gave WHEA errors but was unstable in the last week of it's life.
My first was just unstable with WHEA errors etc.
My 3rd has had one crash 24hrs after installation, I've been trying but I haven't been able to reproduce the issue since, I've only had it for a week.

I've been spending several months researching this, nobody seems to have a definitive answer except lowering timings, upping voltages and trying to get a better CPU. I did hear AMD was aware of an instability issue and were addressing it in the BIOS around Feb/Mar but not sure if that was this specific issue and may not fix all CPUs.

If you haven't already I'd start by updating your BIOS, clearing your CMOS, updating your chipset drivers and all Windows etc.
Things that work for some might not work for others it depends on the CPU I think.

[BNITRO]

OK i want you to do TURN IT OFF. unplug it. troubleshoot time, check the power supply connectors then motherboard then GPU then RAM then NVME, if all is good CHECK your thermal paste under your heatsink.

5800xNitrO 5.1Ghz all cores except one for cooling cycling Modified B550MSI-Gaming-Nitro Modified XFX RX 6800 Modified 37k Passmark Micron PCIE3 Modified NVME 4204Mbs 1Tb-1Tb SanDisk Modified 657Mbs SSD / WD Green Modified 5400 1`40Mbs

3Dmark wont let me BEAT a 3090 and it sucks. I can see the stack trace. its planned.

Micron RAM 4000mhzCL16 with boost to 4200mhz

1000Watt MINER BRAND 80+ GOLD only power supply for $90 with 2 full 92AMP PCIE power RAILS these 6000 series LOVE CLEAR CLEAN POWER WITH NO DATA TRAFFIC FROM THE WALL USING INTERNET THREW POWER METHODS

I'm about to do a refresh on the ram pads to see how they held up to 150Mh with a 6800-116watts and a 590-120watts video output and windows use. FATBOY.

and I can play 4k 150+fps all day long in a 110F room i know I've tested it.

[steveoeditz]

Just wait its about to get worse you will get error 19 which is fatal hardware whea logger error. Having the same issue waiting on the rma from amd and I'm also rma the motherboard to see if there's any issues with it just to make sure both items are checked. This starts out like this and will eventually lead to you having to run a 5900x with cpb and pbo off just to use the system at 3.7 GHZ do yourself a favor RMA it before it doesn't work anymore. 

[Louier]

Yeah I've sent in an rma now, hopefully they find something and don't just send it back broken

[Cmdr-ZiN]

@Louier I dunno if they found something for my first CPU, my store didn't say they said anything, but from my experience and the experience of others they're just replacing them. I haven't heard anyone who got a replacement saying AMD confirmed it was defective.

However I was told I'd get my original CPU back if AMD didn't find a problem. However AMD might just be giving customers the benefit of the doubt.

My second CPU was working perfectly until in the last week it became unstable then could still post but wouldn't boot any OS. It's much nicer when you either have no issues or a clear reproducible issue.

[Gwillakers]

I doubt that AMD has the time, or wants to spend the labor testing the CPU's.

However, Know this:   If AMD does take the time to test these returns, they will do so only under strict specifications.

They will not turn on CPB, or PBO.   They will Not turn on XMP or DOCP. 

They will probably run under a Windows Balanced Power setting.

They will run a Suite of programs.  If it crashes, they will accept the return.

However, your guess is as good as mine as to how long they will wait before accepting the chip as good, and then returning it.

[Cmdr-ZiN]

@Gwillakers I doubt you know what AMD will test or not test. XMP/DOCP is considered a default setting for Intel and many AMD users, also CPUs Boosting is considered part of it's normal function. If you were to claim AMD wouldn't bother testing advertised features then that's a reason never to by AMD.

I don't believe this is how AMD operates, lets try to be more objective.

[Gwillakers]

I never claimed I knew what AMD tested.

However, I still stand by the statement that they will not test with parameters that will go above and beyond specifications.

I believe this is very objective. 

The settings that come with the Motherboards, are put there by the Motherboard Manufacturer, not authorized by AMD.

As far as Max Boost, AMD is referring to the maximum frequency achievable by a single core on the processor running a single threaded bursty workload.

My post was intended to alert the forum, that AMD only has to provide the service specified by specifications, and not qualify processors that are running under conditions for which they were never engineered.

[Cmdr-ZiN]

@Gwillakers CPB is a default setting see https://www.reddit.com/r/Amd/comments/8djs6r/precision_boost_2_vs_xfr2_vs_core_performance/?utm_sour...

As for DOCP/XMP yeah they probably won't test that. After further research this may even void your warranty on some CPUs. Not all though.

However most people recommend XMP and I've never had a issue with it before. If CPU vendors were going to start having issues with it and that was disclosed you'd find people would vote with their wallets.

When I first started using XMP it didn't overclock the CPU, I didn't stop to think this changed when they moved the memory controller to the CPU from the north bridge. However I had an Athlon 64 back in the day it had no issues with XMP and neither have any of my Intel CPUs. Some Intel CPUs have overclocking covered in Warranty.

AMD really shouldn't put 3600Mhz and higher memory on their marketing without a disclaimer that it voids your Warranty. However at the same time AMD shouldn't allow XMP to do any damage.

Still looks like they won't reject a warrant for XMP anyway. I can imagine the outcry if they did.

Anyway this is a moot point because the issues with my first and second CPUs were their without DOCP applied. All default stock settings.

I never even enabled DOCP with my second one, I must of forgot to enable it after initial testing.

I did in my searches find an AMD article saying CPB/Precision Boost requires the latest chipset drivers. Also the chipset drivers appear to contain a stack of fixes and I've not had an issue since the latest update. So maybe that's it.

Or maybe it's just cosmic radiation causing our issues