Pstate OC'ing Issue - Vcore does not reduce when idle?

[Ken429]

The attachment describes what I've done with my Ryzen systems. In both cases, Taichi X370/1700X and the Hero VII/2700X they do not do what I think they should when Overclocked and Idle. Both of them are running the latest greatest BIOS, both are OC'd via Pstate0 and both of them are very stable at 3825 and 4150, respectively. The only issue I have is the 1700X reduces the Vcore when Idle but not the CPU multiplier and the 2700X reduces the CPU multiplier but not the Vcore when idle!

I have post this issue on the ROG and the TweakTown forums but have not found anyone with the same problem or someone with a solution. I cannot believe I'm the only person screwing around with these system configurations in the Ryzen world that has this problem(s)? I know it's not a major issue since both systems tend to cool down to a normal temperature <30C when idle but I have several OC'd Intel 4790K based systems that down clock the CPU and Vcore when I idle so.. I was expecting AMD Ryzen to act the same?

The Hero VII/2700X system does act the way it should when everything is set to default. The BIOS lowers the Vcore and CPU multiplier when idle. The only problem with that, in my opinion, is the system pushes 1.45+Vcore every time it wants to boost the CPU to 4350 on one core. However, As soon as I increase the Pstate0 FID to whatever the system locks the Vcore but at least the BIOS stops trying to run one core at 4350 every time Windows 10 has a request.

Is all this normal for Ryzen BIOS's or is there something I'm missing that needs to be changed?

 

[trents]

I think what you are experiencing is pretty much the norm in the Ryzen world and the variances between the two boards might be accounted for by the fact that you are running different generations of Ryzen CPUs. The P states don't seem to play by the same rules as the Intel platform does, but even there, things have changed most recently. These high voltage spikes with Ryzen are momentary and not a cause for concern. I would just enjoy the powerful system and not get caught up with how you think the P states should behave base on earlier architectures.

 

[Ken429]

Yeah, your probably correct but I have been fighting the Ryzen OC stuff since day two of the 1700X release. At one time many many BIOS upgrades ago the BIOS on the ASRock Taichi X370 MB had it right - it down clocked the CPU multiplier and reduced the Vcore when over clocked like it was supposed to to. Then the new improved BIOS upgrades made that go away! I was hoping that ASUS would eventually get it right on the Hero VII MB but again after several BIOS updates it still is flaky when trying to OC the CPU - at least in my in my opinion.

 

[trents]

Yep, some of those bios upgrades are more downgrades than upgrades. We have recent CPU vulnerabilites (Spectre and

With ASRock motherboards it's easy to roll back to an earlier bios version of the newer ones cause problems. You just do it the same way as you did the upgrade. Motherboards by other manufacturers won't let you roll back but there are third party utilities that serve for that purpose. We have those recent chip vulnerability patches to thank for the way the bios upgrades mess with things. It might be to a good idea to roll back to a pre Spectre/Meltdown version of bios. You should be able to do it with the first generation system you have at least.

 

[Ken429]

Trouble with going back is I would lose the improvements made to handle memory. I have G.Skill FlareX 3200-14 on both machines and like to run the memory faster than 2933 - it allows Handbrake to run a little faster when converting Movies to MP4 files. So...I'll live with what I got and hope that maybe 10 or 12 BIOS updates from now maybe they will finally fix my Ryzen pet peeve. The thing that still bothers me is apparently is at least one person has their 1700X system working correctly using Pstate0 - or at least say it is:

https://rog.asus.com/forum/showthread.php?93319-A-Really-Clear-Manual-P-State-Overclocking-Guide

I was hoping someone else on this forum might have a configuration similar to mine that has the Pstate0 working "correctly" and could shed some light on what I'm missing!

 

[trents]

At some point, the manufacturer will not be producing new updates for that old board. They move one since they can't make money by continuing to support older products.