[TwoEyedJack]

Hello, for the past months I struggle with stability issues on my r7 5800x. Right now I get black screens/restarts even while watching youtube. The history and the solutions that I have tried are many and span a number of months so i will provide a full history at the end of this post if you have the time to read it.

Right now my system consists of a r7 5800x (4month old), msi b550 gaming edge wifi (4month old), gskill ripjaws 2x8gb 3200mhz (bought in late 2018), a Thermaltake smart se 530w (i know its bad (bought in late 2018)) and a be quiet dark rock slim + a 6600xt . I have updated the bios to the latest stable one ( just 4 days ago) and i have only enabled xmp. Yet the system blackscreens and restarts. The timing is weird It happens every week or couple of days and then it stops for a couple of days, when it does happen i get a couple of crashes back to back with varying frequency. The crashes come with the same WHEA errors about Cache Hierarchy Error on pretty much all the APIC ID (most common i suppose is APIC ID 2, 12,14. best cores are 3, 7, 1, 2, 6, 5, 4, 8 ) (look at screenshot for periodicity of the errors ).

Thats pretty much the gist of the issue. Its unstable even at default settings (pbo auto) with just xmp enabled i don't think 3200 is pushing the memory controller, what chould be the cause of this ?



For a more detailed description.
I bought the 5800x and the b550 mobo 5 months ago to upgrade my r5 2600, b450 system. so the ram and the psu is repurposed from a system that was running just fine for about 3 and a half years.

First I didnt knew much about PBO oc so I just enabled it set ppt to 120 , tdc 95, edc 120. and curve optimizer -20 "bad cores" - 15 the two "good" cores and max turbo +50mhz. The system run fine with no crashes for about 2 months. 0 problems in multithreaded applications or single threaded and it frequently boosted to 4850-4900 while gaming. Suddenly 2 months later the crashed started in extreme intensity, meaning the pc would constantly crash back to back even while just trying to browse my files(so i doubt it the usual instability that is observed with pbo one in a blue moon because two months is a long time) .

Then the struggle started, At first i tried core cycler, I run it for many nights (trying different settings per core and different FFTs always for more than 2 iterations of 1hour per core) I found the "stable settings"-nothing extreme best cores at -3 bad cores at -20 and some at -13- that gave me no errors and the crashes stopped.(at the same time i tried using xmp off and even pbo off)

After some weeks of perfect stability the same thing happend, the pc started being extremely unstable. I further increased the curve settings till it was stable for a couple of days/weeks. Then it would once again go into an extreme crashing period where the once perfectly stable system would continuously crash many times a day.

The same pattern of: no crashes for weeks followed by many crashes in a single day remedied by me increasing the curve settings (always run core cycler to check for errors) would continue for many weeks till i reached positive number in some cores to make them "stable".

Now I gave up on curve optimizer i put everything in default except xmp, the system run for a couple of days and then started crashing again so 4 days ago I flashed the newest bios and again 4 days with no crashed and today it crashed twice one in YouTube and once in a game. For a last resort (before trying pbo off and performing an all core traditional oc pr just leaving it with the old boost technology) I enabled curve optimizer all core to positive +5.

I have tried using loadline calibration control for both cpu and soc voltages (soc stays at 1.0 even at prime 95)
I have tinkered with the gpu, performing ddu installations of the drivers various oc settings and I can say that this was not the issue right now it is just running with a power limit +20 setting.

I cant think of anything else to do now, and it has started really exhausting me, i am afraid of using it nowadays. I dont want to admit it to myself but maybe i have a bad cpu and i need to send it back which is a big pain since this is my main computer. I suppose maybe the mobo or the gpu chould be at fault or more possibly that sketchy, old, psu. Or even my installation of windows? Maybe the chipset drivers? Who knows even my house power wiring might be at fault and any power surge in the system from a different defice like an AC or a washing machine affect its.

I waited this long to post it because till now i was overclocking but now we are at stock. I have read many threads on ryzen 5800x issues but i ve never seen this particular behavior.

Thank you for reading my whining post.

Take a look at the periodicity of the errors bellow in case i didn't explained it well. I have colored them in "blocks" keep in mind that every crash grouping is accompanied by me not doing anything for the first few crashes of the grouping and then changing the settings.

 

[Lionvibez]

RMA that CPU if you are still getting those with everything at stock settings and PBO & CO both off.

 

[Hueristic]

Thermaltake smart se 530w

 

[Rei86]

Thermaltake smart se 530w

Probably with the OC and this PSU, I would guess severe degradation to the chip might have taken place on the worst case.

 

[TwoEyedJack]

Probably with the OC and this PSU, I would guess severe degradation to the chip might have taken place on the worst case.

Severe degradation in 2 months because the psu is of bad quality? Regardless of the fact that it was working for many years a ryzen 2600 with no "degradation" whatsoever? The 5800x is pulling at max 120w just like the 2600 was. And besides cpu don't degrade like this.

 

[TwoEyedJack]

Thermaltake smart se 530w

Yes funny and bad. But chould you point at how this bad psu would affect it? Power wasn't the issue. Maybe unstable voltage delivery?

 

[TwoEyedJack]

RMA that CPU if you are still getting those with everything at stock settings and PBO & CO both off.

I understand that's the logical action to do. It's just a pain to deal with amd themselves and not the retailer because no computer shop would understand this issue.
Could the os have anything to do with it? Or the psu? I wanted to mark out these factors before I proceed with replacements

 

[affxct]

I understand that's the logical action to do. It's just a pain to deal with amd themselves and not the retailer because no computer shop would understand this issue.
Could the os have anything to do with it? Or the psu? I wanted to mark out these factors before I proceed with replacements

Inform them that you're having crashes and that your testing indicates it's a CPU issue and that you haven't done anything weird to it so you'd like a replacement.

 

[The_King]

Run your ram with XMP disabled and see if you still get the same issue.

Some boards dont given enough voltage for XMP try increase that by 0.1V or 0.2V.
Post your ZEN timings.

Has far has CO goes it does not matter how low you think you running the good or bad core. It can crash anyway. Infact i get random reboots with "low CO" like -14 on core 1 but not with -29 on the same core.

OCing your CPU on a bad PSU is just wrong.

 

[ku4eto]

With Ryzen, black screen crashes are basically issue with CPU-RAM communication.
In my case, that was a 3200Mhz RAM with a R7 1700 CPU. It was simply not stable.

Now, seems like you are running 3200Mhz on a 5800X, which SHOULD be okay. As someone else mentioned, try disabling the XMP, run the RAM at 2400Mhz and see if it fails again.

If not, set manually the voltage after enabling the XMP profile, along with the DIMM frequency.

 

[Shawnb99]

If you know the PSU is a POS why are you still using it? Sure it could be something else causing the issues but you already know the PSU is crap so why keep using it?
I’ll never understand how people can spend on everything else yet cheap out in the PSU.

 

[TwoEyedJack]

Run your ram with XMP disabled and see if you still get the same issue.

Some boards dont given enough voltage for XMP try increase that by 0.1V or 0.2V.
Post your ZEN timings.

Has far has CO goes it does not matter how low you think you running the good or bad core. It can crash anyway. Infact i get random reboots with "low CO" like -14 on core 1 but not with -29 on the same core.

OCing your CPU on a bad PSU is just wrong.

These are the timing. There was no difference between them using xmp profile 1 and profile 2 (same mhz).Hwinfo reports dram of 1.372V average and max, minimum its 1.368. I don't think that its a good idea to increase it by 0.1 or 0.2V, I have never seen anyone running them above 1.45 max 1.5 for benchmarks.. Do you see anything wrong with timings or other voltages.? Infinity fabric is also at stock running at 1600. I haven't tried SoC to 1.1I ve only messed with lcc but i wouldn't think that 1.0v is not enough for 1600 IF

With Ryzen, black screen crashes are basically issue with CPU-RAM communication.
In my case, that was a 3200Mhz RAM with a R7 1700 CPU. It was simply not stable.

Now, seems like you are running 3200Mhz on a 5800X, which SHOULD be okay. As someone else mentioned, try disabling the XMP, run the RAM at 2400Mhz and see if it fails again.

If not, set manually the voltage after enabling the XMP profile, along with the DIMM frequency.

The specific kit is listed in msi's compatibility list for this specific board and cpu combo so I fail to understand why it wouldn't work. If it was memory issue wouldn't i get a different WHEA error ? I have tried no xmp in the past and it still crashed. But i will try it again. About manual settings i am not sure what to dial in the more.. exotic timings. Also is there any stress test able to bring out a problem like this? I have tried memtest but i didn't see any errors

 

[The_King]

VSOC below 1V seems low to me. Not sure if that because of the PSU. Low VSOC can cause crashes for sure

I would up that to 1.05V-1.08V and see if that makes a difference.

 

[TwoEyedJack]

If you know the PSU is a POS why are you still using it? Sure it could be something else causing the issues but you already know the PSU is crap so why keep using it?
I’ll never understand how people can spend on everything else yet cheap out in the PSU.

It was working for me. When I upgraded cpu,mbo and gpu i pretty much stayed at the same wattage tire components. The computer never pulls more than 250 watts, except if i run furmark and prime at the same time, which is 50% of the rated wattage.
But I suppose you are right, regardless of this matter I should change it. Will do so in the following days.

 

[TwoEyedJack]

VSOC below 1V seems low to me. Not sure if that because of the PSU. Lowe VSOC can cause crashes.

I would up that to 1.05V-1.08V and see if that makes a difference.

The software reported 12v, 5v and 3.3v values are on the limit of the ATX Standard. Indeed it might affect it. Some times I've seen vsoc down to the 0.9 with previous bios. Should I enable llc and and set overcurrent protection to enhanced ? I've read conflict things about them, or just up the soc to 1.05

 

[The_King]

View attachment 2566615
The software reported 12v, 5v and 3.3v values are on the limit of the ATX Standard. Indeed it might affect it. Some times I've seen vsoc down to the 0.9 with previous bios. Should I enable llc and and set overcurrent protection to enhanced ? I've read conflict things about them, or just up the soc to 1.05

I dont use LLC so I cant recommend it.

But I know for sure a low VSOC can cause black screens on ZEN 3.
Try setting it to 1.05V

 

[Rei86]

Severe degradation in 2 months because the psu is of bad quality? Regardless of the fact that it was working for many years a ryzen 2600 with no "degradation" whatsoever? The 5800x is pulling at max 120w just like the 2600 was. And besides cpu don't degrade like this.

I said that's the worst case scenario

From what I can go off from what you've posted..

1. You have a bronze quality PSU from 2018 that's been used since till now, that's a four year old PSU that you've already stated you don't even like. Its rated at 530w
2. You have a 5800X and you ran it overclocked, a stock 5800X already pulls over 100w we already know
3. You have a 6600XT which can pull over 160w alone stock

So worst case its looking like you'll need to RMA that CPU and pick up replacement PSU.

 

[Slaughtahouse]

Have you tried disabling XMP yet? XMP isn't guaranteed and memory is usually the pain with Zen from my experience.

If your system can't run stock memory and you're still having WHEA errors - go through the pain and RMA the CPU. If errors go away, its on to the next troubleshooting step.

Beyond that - regardless of stability, as others mentioned, replace the PSU. Just don't be surprised if you do swap it, you still have stability issues with your current configuration. We just don't want to see your system burst into flames lol.

FWIW: I did research that unit. It's not terrible... Thermaltake Smart Series 530 W Power Supply Review - Overclockers but there are much better options out there without clumps of solder lose. Yes, 38A on the 12v rail is more than enough for a 5800X + 6600XT but there is more to a PSU than just power output.

Spoiler: Good vs. Bad Power Supply Differences Explained by PSU Expert Jonny Guru

 

[o1dschoo1]

Set vsoc to 1.15 and test. Yes 1.15 is a little high but still under max safe voltage. If that doesn't fix your issue rma it.
Also verify what your fclk is running also

 

[GRABibus]

Hello, for the past months I struggle with stability issues on my r7 5800x. Right now I get black screens/restarts even while watching youtube. The history and the solutions that I have tried are many and span a number of months so i will provide a full history at the end of this post if you have the time to read it.

Right now my system consists of a r7 5800x (4month old), msi b550 gaming edge wifi (4month old), gskill ripjaws 2x8gb 3200mhz (bought in late 2018), a Thermaltake smart se 530w (i know its bad (bought in late 2018)) and a be quiet dark rock slim + a 6600xt . I have updated the bios to the latest stable one ( just 4 days ago) and i have only enabled xmp. Yet the system blackscreens and restarts. The timing is weird It happens every week or couple of days and then it stops for a couple of days, when it does happen i get a couple of crashes back to back with varying frequency. The crashes come with the same WHEA errors about Cache Hierarchy Error on pretty much all the APIC ID (most common i suppose is APIC ID 2, 12,14. best cores are 3, 7, 1, 2, 6, 5, 4, 8 ) (look at screenshot for periodicity of the errors ).

Thats pretty much the gist of the issue. Its unstable even at default settings (pbo auto) with just xmp enabled i don't think 3200 is pushing the memory controller, what chould be the cause of this ?



For a more detailed description.
I bought the 5800x and the b550 mobo 5 months ago to upgrade my r5 2600, b450 system. so the ram and the psu is repurposed from a system that was running just fine for about 3 and a half years.

First I didnt knew much about PBO oc so I just enabled it set ppt to 120 , tdc 95, edc 120. and curve optimizer -20 "bad cores" - 15 the two "good" cores and max turbo +50mhz. The system run fine with no crashes for about 2 months. 0 problems in multithreaded applications or single threaded and it frequently boosted to 4850-4900 while gaming. Suddenly 2 months later the crashed started in extreme intensity, meaning the pc would constantly crash back to back even while just trying to browse my files(so i doubt it the usual instability that is observed with pbo one in a blue moon because two months is a long time) .

Then the struggle started, At first i tried core cycler, I run it for many nights (trying different settings per core and different FFTs always for more than 2 iterations of 1hour per core) I found the "stable settings"-nothing extreme best cores at -3 bad cores at -20 and some at -13- that gave me no errors and the crashes stopped.(at the same time i tried using xmp off and even pbo off)

After some weeks of perfect stability the same thing happend, the pc started being extremely unstable. I further increased the curve settings till it was stable for a couple of days/weeks. Then it would once again go into an extreme crashing period where the once perfectly stable system would continuously crash many times a day.

The same pattern of: no crashes for weeks followed by many crashes in a single day remedied by me increasing the curve settings (always run core cycler to check for errors) would continue for many weeks till i reached positive number in some cores to make them "stable".

Now I gave up on curve optimizer i put everything in default except xmp, the system run for a couple of days and then started crashing again so 4 days ago I flashed the newest bios and again 4 days with no crashed and today it crashed twice one in YouTube and once in a game. For a last resort (before trying pbo off and performing an all core traditional oc pr just leaving it with the old boost technology) I enabled curve optimizer all core to positive +5.

I have tried using loadline calibration control for both cpu and soc voltages (soc stays at 1.0 even at prime 95)
I have tinkered with the gpu, performing ddu installations of the drivers various oc settings and I can say that this was not the issue right now it is just running with a power limit +20 setting.

I cant think of anything else to do now, and it has started really exhausting me, i am afraid of using it nowadays. I dont want to admit it to myself but maybe i have a bad cpu and i need to send it back which is a big pain since this is my main computer. I suppose maybe the mobo or the gpu chould be at fault or more possibly that sketchy, old, psu. Or even my installation of windows? Maybe the chipset drivers? Who knows even my house power wiring might be at fault and any power surge in the system from a different defice like an AC or a washing machine affect its.

I waited this long to post it because till now i was overclocking but now we are at stock. I have read many threads on ryzen 5800x issues but i ve never seen this particular behavior.

Thank you for reading my whining post.

Take a look at the periodicity of the errors bellow in case i didn't explained it well. I have colored them in "blocks" keep in mind that every crash grouping is accompanied by me not doing anything for the first few crashes of the grouping and then changing the settings.
View attachment 2566599


View attachment 2566600

View attachment 2566598

Don’t spend much more time to find a solution.
RMA ! 😊