Weird memory timings with Ryzen 5 5600X + freezing issues?

[VuurVOS]

Is that on a weekly, monthly or yearly basis?

 

[cst1992]

Monthly, maybe? It's not very frequent.

 

[VuurVOS]

Monthly sounds frequent to me. In my region it is only once a couple of years.

Do you have a different power supply which you can put in the system?

 

[Ketxxx]

@cst1992 That's a good thing, it conclusively says you shouldn't be having any memory problems whatsoever anymore so we should be able to put memory related problems to bed now, same for the IMC. If one of your CPU cores is having problems then locking the CPU to a flat 3.7GHz should expose that, what voltage is your CPU currently using when it is fully loaded? Degradation could be the root cause if the mainboard has been throwing way too high of a voltage at the CPU when PBO was enabled. I'd never trust PBO with voltages because of how manufacturers can manipulate it and because what voltage your CPU gets with PBO can vary from one firmware to the next just depending on what was changed in the code. You don't want that kind of uncertainty at the best of times let alone when trying to diagnose an issue.

 

[A Computer Guy]

@cst1992 That's a good thing, it conclusively says you shouldn't be having any memory problems whatsoever anymore so we should be able to put memory related problems to bed now, same for the IMC. If one of your CPU cores is having problems then locking the CPU to a flat 3.7GHz should expose that, what voltage is your CPU currently using when it is fully loaded? Degradation could be the root cause if the mainboard has been throwing way too high of a voltage at the CPU when PBO was enabled. I'd never trust PBO with voltages because of how manufacturers can manipulate it and because what voltage your CPU gets with PBO can vary from one firmware to the next just depending on what was changed in the code. You don't want that kind of uncertainty at the best of times let alone when trying to diagnose an issue.

My understanding with PBO on AM4 is as long as you don't disable the CPU's voltage control (including keeping scalar to 1x) you can't damage the chip with raising PBO limits alone. (vrm issues aside)

 

[Ketxxx]

My understanding with PBO on AM4 is as long as you don't disable the CPU's voltage control (including keeping scalar to 1x) you can't damage the chip with raising PBO limits alone. (vrm issues aside)

Correct, but on heavier CPU loads PBO does have a habit of throwing extra voltage around even if the CPU doesn't need it because the algorithm can't account for each specific CPU. That extra voltage does also tend to vary depending on what board you have, VRM quality, and what the firmware devs of any given manufacturer decided they wanted to do. The chain can potentially go something like this; Mid\high end CPU > sub-par VRMs (read: not necessarily bad, just not really adequate enough for ideal conditions) > more voltage pushed > firmware devs try to compensate > +/- voltage pushed > variable conditions > CPU receives more voltage than necessary. It's that sort of chain of events why even my own system with PBO will ramp the CPU up to 4.4GHz but do so while giving it 1.5v+ (around 1.53v is the highest I've seen if I remember correctly) while with manual tuning it didn't take me long to find out the CPU is absolutely stable @ 4.4GHz with 1.306v, that's a lot of extra voltage PBO is pushing for no reason and if there's one thing the X570 Tomahawk does really quite well is the component choice for VRMs so the only problem here is an overzealous PBO.

Things like this are also why a while ago Gigabyte released a firmware that killed CPUs, the firmware got pulled when it was discovered naturally but it just highlights another problem - manufacturers don't test enough (or at all) prior to releasing the end users are the testing phase and if some hardware smokes in the process... well, good luck getting the manufacturer to give you the money to replace it all you can do is hope that component can be RMA'd with the manufacturer of it.

I honestly would have thought AMD would have learned more for their PBO algorithm and made more improvements considering they must use a derivative of it for voltages in their GPUs. Polaris cards particularly were guilty of throwing way too much voltage out silicone that only needed like 1.08v for optimal frequency (which still far exceeded manufacturer specified frequency) still had like 1.2/1.25v thrown at it which naturally lead to excessive heat and power draw as well as unnecessary degradation of the GPU over time - which I have first hand experience of as part of the lengthy article I wrote HERE I actually kept one card in another system just letting it do whatever the firmware/driver defined. It died a few months ago because a VRM phase blew and it's not like it was a cheap Polaris varient either, RX580 Sapphire Nitro+. The only other Polaris card I have these days is an RX470 Nitro+, that card I modified from day one so has never had any excessive abuse and it is still very much alive.

 

[cst1992]

Monthly sounds frequent to me. In my region it is only once a couple of years.

Do you have a different power supply which you can put in the system?

I do, but it's even older than this one and only 430W.

If one of your CPU cores is having problems then locking the CPU to a flat 3.7GHz should expose that, what voltage is your CPU currently using when it is fully loaded? Degradation could be the root cause if the mainboard has been throwing way too high of a voltage at the CPU when PBO was enabled.

Was PBO enabled? I didn't change any settings related to that in the BIOS, only in Ryzen Master - and that too only when I was doing some workstation task like converting a video. I don't even run Ryzen Master by default. For everyday tasks it'd be a pretty safe assumption that the CPU was running on manufacturer-specified values.

As for voltage - I'm looking at VID and that is set at 1.1V at idle, and 1.0687V at load.
It's also subject to minor fluctuations such as going to 1.0938V at idle and 1.0750V at load.

Does a stock 5600X consume 33W at idle though? That's 50% of TDP.

 

[VuurVOS]

I do, but it's even older than this one and only 430W.


Was PBO enabled? I didn't change any settings related to that in the BIOS, only in Ryzen Master - and that too only when I was doing some workstation task like converting a video. I don't even run Ryzen Master by default. For everyday tasks it'd be a pretty safe assumption that the CPU was running on manufacturer-specified values.

As for voltage - I'm looking at VID and that is set at 1.1V at idle, and 1.0687V at load.
It's also subject to minor fluctuations such as going to 1.0938V at idle and 1.0750V at load.
View attachment 323444View attachment 323443
Does a stock 5600X consume 33W at idle though? That's 50% of TDP.

430W is a "tight" fit for your system under load.

I think your CPU vcore LLC is still on auto in bios. Please put it on 4 so the voltage drop is less.

 

[cst1992]

I think your CPU vcore LLC is still on auto in bios. Please put it on 4 so the voltage drop is less.

Done. Set to level 4 and now VID goes to 1.0938V under max load. Also consumes about 3W more power (maybe the point of the Auto setting?)

 

[oobymach]

Re: bclk - I'm looking at the BIOS settings now - there is no dropdown. There's only a field and it's inputted 100.0000 by default with a note that you can set it 96.0 - 118.0.

If you delete the bclk numbers and hit enter does it go to auto or simply reset to 100.00?

That's what I had to do to make mine auto after manually setting it.

 

[wNotyarD]

If you delete the numbers and hit enter does it go to auto or simply reset to 100.00?

I'll add to this that maybe you must type Auto on the field and hit Enter. I know for a fact it's a possibility on Gigabyte's BIOS.

 

[cst1992]

FINALLY, another "different error" moment. Seriously, I feel like throwing a party right now.
I was able to get a kernel dump (was getting volmgr error for complete dumps), and it says the following (via WhoCrashed):

If the CPU isn't downright bad, maybe the issue can be resolved by replacing this driver file. Thoughts?
Personally I'm happy that I'm at least getting crash dumps now, but this doesn't really tell us anything new.

Also, this thread advises setting a value in registry to 4: https://answers.microsoft.com/en-us...mdppmsys/941dd57b-1bde-453d-a9d9-278942ff38b5

Navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\amdppm
Change Start to have a value of 4.

It capped the CPU to 3 GHz, but it (apparently) solved his issue.
Mine is currently set to 3.

 

[A Computer Guy]

If the CPU isn't downright bad, maybe the issue can be resolved by replacing this driver file. Thoughts?

Well I guess it can't hurt to reinstall the AMD chipset driver. (assuming that is where that file came from, try the most recent one)

 

[cst1992]

Will do. I think I was actually using the latest drivers straight from AMD's website, but may have removed them to install the ones on the motherboard page, to try and use what Asus were recommending.
Reference: https://www.techpowerup.com/forums/...n-5-5600x-freezing-issues.315089/post-5146254

 

[VuurVOS]

Maybe it is also wise to reseat the processor a couple of times. (Open socket, remove processor, put it back, close socket and repeat)

 

[cst1992]

I've reinstalled the chipset driver. Motherboard's website was using 4.07.13.2243, and the latest version is 5.08.02.027.

 

[Ketxxx]

I do, but it's even older than this one and only 430W.


Was PBO enabled? I didn't change any settings related to that in the BIOS, only in Ryzen Master - and that too only when I was doing some workstation task like converting a video. I don't even run Ryzen Master by default. For everyday tasks it'd be a pretty safe assumption that the CPU was running on manufacturer-specified values.

As for voltage - I'm looking at VID and that is set at 1.1V at idle, and 1.0687V at load.
It's also subject to minor fluctuations such as going to 1.0938V at idle and 1.0750V at load.
View attachment 323444View attachment 323443
Does a stock 5600X consume 33W at idle though? That's 50% of TDP.

I would say most things look fairly normal there although 1.068v is rather a long way off of VID 1.1v, depending on system workload if the voltage doesn't ramp up quick enough I could see that causing some stability issues. I'm not sure exactly how voltages work on your board but if it's like the Tomahawk I'd adjust VID to 1.15v or dial in a manual voltage of 1.2v. Experimenting with LLC and seeing which setting holds closest to the voltage you set wouldn't be a bad idea either to ensure the supplied voltage is as stable as possible. I also wouldn't use Ryzen Master and just do everything within the boards firmware to get rid of as many unnecessary layers and potential bugs\compatibility issues as possible. I've said it for years, Windows OC software is one of the worst scourges that has ever existed.

EDIT: Seen your last post, that pretty much confirms what I've been thinking and saying. Reseat the CPU, make sure no TIM has got on the pins or in the socket, ensure the mounting on the CPU is good. If all that is fine combined with the suggested voltage change then it's looking more likely the CPU is iffy. Still other things to test to be sure, such as if you can't get another PSU throw in a cheap low power GPU to try and rule out PSU problems.

 

[cst1992]

@VuurVOS @Ketxxx I just realized one thing after reinstalling the chipset driver.

When I start installing drivers after a new system installation, I usually do so after going to Device Manager, looking at the list of devices, then going to the manufacturer's page for each device, downloading the drivers, and then installing them via installers and/or loading via Device Manager as appropriate.

What I realized is that the first time I installed Windows on this new build back in September, I forgot to install the chipset driver on this system for over a month. But the system was unstable due to the incompatible RAM and stuttering so all of my attention at the time was focused on that. Plus, I was using the original BIOS that was on this board (0309) so any freezes were attributed to it being an old BIOS.

If the system keeps on freezing/BSODing due to the amdppm.sys file which is part of the chipset drivers, maybe I could just... not use them?
I mean, it worked for a month without them, right?

 

[Ketxxx]

@cst1992 AMDPPM is the Primary Processor Module, so while that system driver could have a bug in it causing problems it is more likely the errors are happening because the CPU itself has a fault. Before jumping to that conclusion though I'd start by using a much older set of chipset drivers, say from 6-12 months ago but that would mean yet another format to ensure no remnants from the current chipset drivers stick around like stagnant air.

 

[A Computer Guy]

@VuurVOS @Ketxxx I just realized one thing after reinstalling the chipset driver.

When I start installing drivers after a new system installation, I usually do so after going to Device Manager, looking at the list of devices, then going to the manufacturer's page for each device, downloading the drivers, and then installing them via installers and/or loading via Device Manager as appropriate.

What I realized is that the first time I installed Windows on this new build back in September, I forgot to install the chipset driver on this system for over a month. But the system was unstable due to the incompatible RAM and stuttering so all of my attention at the time was focused on that. Plus, I was using the original BIOS that was on this board (0309) so any freezes were attributed to it being an old BIOS.

If the system keeps on freezing/BSODing due to the amdppm.sys file which is part of the chipset drivers, maybe I could just... not use them?
I mean, it worked for a month without them, right?

The thing is your system shouldn't be freezing/BSODing at all. It's in a unreliable state for some (possibly multiple reasons) why use an unreliable computer that's going to crash when you need it?
I'm almost inclined to think you might need to start swapping out other parts at this point.

I'm using the latest chipset driver 5.08.02.027 on all my AMD systems (Win 10) and not having these issues at all. I don't think it's the driver unless on your disk it was damaged.

 

[cst1992]

this thread advises setting a value in registry to 4: https://answers.microsoft.com/en-us...mdppmsys/941dd57b-1bde-453d-a9d9-278942ff38b5
It capped the CPU to 3 GHz, but it (apparently) solved his issue.
Mine is currently set to 3.

okay, what about this?

 

[A Computer Guy]

okay, what about this?

You ought not have to do anything special with registry. Can you get a CPU replacement under RMA or AMD warranty?

 

[cst1992]

l'll have to check. For now I have found a technician who might have a spare for testing purposes, we'll see how that goes.

 

[A Computer Guy]

l'll have to check. For now I have found a technician who might have a spare for testing purposes, we'll see how that goes.

Also VurrVOS suggestion for swapping out the PSU seems reasonable as well. You might try that first.

 

[cst1992]

I'm using the latest chipset driver 5.08.02.027 on all my AMD systems (Win 10) and not having these issues at all. I don't think it's the driver unless on your disk it was damaged.

There's speculation online that the problem could have been Windows 11 itself.
Who knows.