Question Question about PBO limits on AM5 Asrock Motherboards

[theinfamoustoast]

Mobo: PG Riptide b650m
CPU: 7700x -15 offset

All I want to do is lower the PPT limit on the 7700x, however when I set PBO limits to 'manual' the other TDC/EDC values are set to 0.

I roughly know what to set my PPT to: 85000mV / 85W but the other values are set to 0 when I go to manual settings.

So my question: Does the value 0 on PPT/TDC/EDC = Default? Or would I have to set those to their correct value too when going manual? I only wanted to change my PPT.

 

[Darkbreeze]

I THINK (I'd have to dig into it and check my Ryzen system in my garage, which might not be helpful since it's a X570 board and might not be relevant to your B650 configuration) that when you manually configure the PBO limits you have to manually configure ALL of them.

More importantly though, why are you trying to manually adjust this limit? PBO is precisely tuned by each board manufacturer for that specific board, so the idea that you (Well, not YOU you, just, most people in general) are going to be able to more precisely or efficiently tune the profile than engineers with decades of experience in designing and configuring computing architectures is not realistic. Yes, they allow it, but you can also set your memory latency to like 10, and for DDR4 or DDR5 that too is just ridiculous and is never going to work or even POST. Just because you CAN do a thing, doesn't mean you SHOULD do a thing.

So unless you have a really good reason, backed by some form of substantial proof or evidence that there is a tangible benefit from doing it, provided by somebody known and respected in this sort of tuning endeavor, I'd say to either simply leave it alone at the factory designated configuration for PBO, or simply manually configure your overclock and basically everybody with any overclocking experience has plainly said that for 99.99% of users, since about Zen 2, there is realistically not much chance of getting a better overclock than what PBO already provides.

But certainly knowing WHY you are wanting to do this, might have some bearing on that? If it's because you're seeing overheating or throttle behavior, then the obvious answer is simply the VRM configuration on that board is not good enough or your cooling solution is not good enough, or both. In which case you should simply disable PBO until such time as you can improve your case and CPU cooler cooling configuration or get a better board.

 

[theinfamoustoast]

HWinfo64 showed 'CPU DIE - Average' 117c - 125c on two occasions with SOC and other voltages being +2V or higher.
OCCT monitoring software showed 'CPU DIE - Average' at 165c
Yesterday HWinfo64 showed my 'CPU DIE' - at 200+c

it plays games fine and with stress tests like Cinebench and Furmark being okay, but sometimes these crazy values show up on multiple monitoring software showing ridiculous temperatures and voltage spikes on the CPU. This has happened twice in the past few days and I've been trying to replicate these crazy results. I don't know exactly what is doing, sometimes using heavy CPU applications, when I'm playing cs:go and other times when I'm idle on Windows.

Apart from XMP, I have only set CPU max junction temp from 95c to 90c. I had PBO set to 'Enabled' when this first happened, then I put it back to 'Auto' afterward, seeing that's the only change I've done CPU-wise and it still caused this result. Updating the Bios did not seem to help. - I have now switched from XMP to EXPO as the timings are better and set an offset on the CPU -15. Not that this should fix the problem but just because I can so why not do it.

Contact with the Dark Rock Pro 4 and 7700x seem to be ok, as I did it months ago and had no issues in the past. The airflow within the PC is also quite high.

Images down below of the CPU/Motherboard and i-GPU from HW64 (high CPU temp)

https://drive.google.com/file/d/1CAFB2UgNhHyUReRW-MuqiN0i-G98ts7r/view

https://drive.google.com/file/d/1xda43j3OsEk2wYX5HF1xPk9NtpWkZZw4/view

https://drive.google.com/file/d/1UUbfvyELzbP_TR8sJgII0_lQX0sXY339/view

I am at the latest update with both monitoring softwares and generally anything related to my PC - GPU, MOBO, CPU ETC.

I've seen a lot of people talking about this to this day, even with staff members on HWinfo64 claiming it's not a bug within the software and a problem with AMD and the sensors in the CPU supposedly.

"This is clearly an erratum in CPU/SMU and AMD should look at that and fix it."
"Those spikes (which seem to be exactly double of what it should be) are in fact invalid/misreporting. So no damage can happen because of this, it's just a misreporting bug that should be fixed by AMD. It should be possible to fix it a BIOS update once they locate the bug." - (Martin HWiNFO Staff member)

Source: https://www.hwinfo.com/forum/threads/hwinfo64-cpu-die-average-spike.8044/page-9

I've emailed both Asrock and AMD about this issue.

Asrock supposedly set up a test bench with my mobo and CPU but could not replicate the results I was having, but did agree with what HWinfo said which was a sensor problem, and that the next AGESA / Bios could fix it.

AMD's response was more like a robot talking to me until I showed them some of my information and after a few emails they just told me to claim a warranty as it seemed like a sensor problem:

"I see that your system is up to date with all the drivers and BIOS. This seems like a sensor issue. I suggest you claim warranty for the CPU using the below link."

So to answer your question, I'm just trying to fix the problem myself but I don't know what to do to replicate the issue. I can run Aida64 Extreme for a full hour with everything ticked with no problems and or crazy temperatures or run a similar thing in OCCT with no errors.

It seems impossible to replicate this issue so I thought of lowering the Watts and trying 'eco mode' that people have been doing to their Ryzen 7000 chips but setting it manually rather than using one of the presets from Asrock. Although this should not fix the problem I'm just clutching at straws here. The last thing I want to do I send this CPU or Motherboard back for an RMA/Warranty as no one wants to deal with that.

Sorry for this being so long lol, thank you for reading.

 

[Darkbreeze]

So, I know you don't "want" to deal with an RMA, but if AMD says they are willing to warranty it, I'd probably recommend you do that. And the reason is, they likely have already known about this, probably have newer stepping models that don't have this problem and even if no actual "damage" is being caused (Which is in fact an assumption, especially since Asrock has stated they couldn't replicate the problem) it might still be affecting performance. It seems very strange to me though that either AMD or ASRock would say this, because generally speaking the motherboard controls all those voltages, not the CPU.

Granted, this is an architecture that I haven't had any hands on experience with yet, the latest AMD architecture I have worked with are all Zen 3 models and I know Zen 4 is moderately different, but it isn't THAT different that we're talking apples vs torque wrenches.

The thing that gives me pause on my beliefs though is if Martin told you that it's a sensor issue, and I know Martin (Not personally in real life, just from having interacted with him many times before on HWinfo issues), then I'd tend to believe him. What really matters though is exactly WHERE the sensors in question are located and what is doing the reporting because again, that is usually the motherboard. And if ASRock has said that a BIOS update should fix it that would further indicate a board issue. I think I'd be having a discussion with them about some kind of guarantee that if a BIOS update doesn't fix it, that they will replace or refund the board so you can get something else because as of yet I have not seen anybody else with this issue so I'm tentatively going to assume that this is motherboard model or manufacturer specific (Meaning ASRock) unless somebody can show me widespread evidence that other motherboards and 7000 series CPUs are exhibiting this same behavior, because so far you are the only one I've come across with this issue.

Which isn't to say there aren't others, but again, it might be only a certain run of boards or CPUs, or limited to a certain model of board, or it even might just be due to specific BIOS releases which again goes back to ASRock.

But in the meantime, I don't think you can manually ONLY change that value, without also manually changing the other values as well. I would think that would be something that Martin or ASRock could probably tell you for sure, however I can also look into this when I get a chance but truly I think that is a very small band aid for a much larger issue and you'd be a lot better off if you just bite the bullet and make them fix the problem correctly with a replacement board, replacement CPU or refund. Or at least a promise that a newer BIOS release is forthcoming that WILL correct the issue.

 

[theinfamoustoast]

Thank you for replying, I will try and tinker around again but I think I will wait for the new AGESA update, as Asus motherboards seemed to receive it.
If that fails, I'll start talking to AMD and or Asrock about a potential RMA. Thanks again.

 

[Darkbreeze]

For sure man. Let me know what happens. Good luck to you going forward.

 

[TerryLaze]

Apart from XMP, I have only set CPU max junction temp from 95c to 90c. I had PBO set to 'Enabled' when this first happened, then I put it back to 'Auto' afterward, seeing that's the only change I've done CPU-wise and it still caused this result.

Just saying but pbO is overclocking(so is xmp) , if you just want lower power/temps just disable/off pbo and have just pb, so the normal boost behavior and not the override.

 

[Darkbreeze]

You can't "disable" something that it doesn't have. And A320 boards don't support PBO, so it doesn't have it.