BIOS 3.00 Pstate OC'ing?

These are the only BIOS settings from the Defaults I changed or may have changed back to the Defaults along the way:
Overclock Mode - Auto
CPU Frequency and Voltage Change - Auto
Load XMP Setting - XMP 2.0 Profile1 (EVGA 3000-15 Memory)
Voltage Mode - Stable
CPU Core Voltage - Auto
CPU Load-Line Calibration - Level 2
VDDCR Load-Line Calibration - Level 2
Global C-State Control - Disable (To make changes to Pstate0 stick)
Global C-State Control - Enabled (After making changes to Pstate0)
Pstate1 - Auto
Pstate2 - Auto
Pstate3 = Disabled

Hope this helps, but I don't think the BIOS settings have much todo with the AMD Ryzen Power Plan. You might want to rerun the Ryzen chipset drivers from the site mentioned above.


Edited by Ken429 - 07 Aug 2017 at 5:25pm

I don't know what else I can offer. Every setting I have played with is in this thread. You might try the old fallback position and CLR CMOS, Load Defaults and go from there.

No, I just leave it Disabled most of time since I forget to Enable it. Don't know what difference it makes.

There is no magic in PState over clocking, meaning it will be no different or better than using the manual settings in OC Tweaker. Frankly all the higher (lower speed) PStates (1, 2, etc) are only useful for down-clocking the core speeds under low CPU loads. If we use the Ryzen "Balanced" power plan, you won't get any down-clocking of the cores, since the Minimum Processor State is set to at least 95% if I recall correctly. The Reset Defaults option for the Ryzen Balanced plan does not work.

So with the default settings of the Ryzen Balanced power plan, supplied by AMD, we cannot use the lower PState core speeds. Also, where is AMD's guide for using PStates with Ryzen processors?

If we go into a PState form, and just begin changing things without understanding what they do, we can only blame ourselves if the results are strange or failure. I have not spent enough time studying PStates to describe them appropriately, but if we need to ask what hexadecimal value to use for the VCore, then we are just stumbling around in the dark and cookbook over clocking.

After changing to the Windows Balanced plan, or setting the Minimum Processor State in the Ryzen plan to a low value (I use 5%), I can get the core speeds to drop to 2,200MHz. I also have CnQ enabled, and C6 enabled. My OC is 3,925MHz, to keep it above 3,900MHz with the drooping BCLK. I have PState0 set to Custom, PState1 set to Auto, and PState2 set to Custom. I changed nothing else on the PState forms.

The hierarchy of PStates is PState0 is the highest processor speed, all the other are simply lower core speeds. Setting the VCore with a Hexadecimal value without any control over the VCore mode (ie, fixed, offset, etc) is not entirely useful, and causes us to use the OC Tweaker screen to set the VCore mode.

The 2,200MHz core speed is PState2, since as we know PState0 is the same as our manual OC in OC Tweaker, it is mirrored in PState0. Or at least it is with the UEFI (3.00 and earlier versions) for my X370 Killer SLI/ac board. I cannot comment on how the X370 Taichi board's UEFI behaves, I don't have that board.

As you know now, enabling a PState is done by setting it to Custom. Skipping or ignoring a PState is done by setting it to Auto. Disabling a PState disables all the others below it. Since the PState0 is the default processor speed, OR the OC speed if you OC, be sure you verify that the settings in OC Tweaker match those of PState0. With my board's UEFI, the OC Tweaker settings are automatically transferred to PState0. But if I modify the PState0 form, the changes are NOT transferred to the OC Tweaker screen. Whether or not that should happen, I'm not sure yet.

Ken429 wrote: parsec, What you stated above don't work for me. If I use the OC Tweaker method to OC @3.8 the system does not down clock or drop the voltage when Windows settles down! Also, the "Minimum Processor State" does not show up under "Processor Power Management" in the AMD Ryzen Balanced Power Plan! So I could not go in and change it to your 5%? However, if I change PstateO to Custom Pstate1 to Auto and Pstate2 to Custom everything works as advertised even the Minimun Processor State is available. So...the key to down clocking was the changes made to the Pstate settings. I only stumbled into the solution by going to the Pstate method based on several different postings. Using the OC Tweaker does the hex conversion automatically in the Pstates. I agree much simpler although once you understand the Hex and how it's used it ain't that complicated. Now, what about my original questions concerning the ASRock A-Tuning- division by zero error, ASRock Restart to UEFI - hanging the system and the Windows Restart - skipping the BIOS screen? Am I the only guy with this issues or is there something I'm doing wrong? It seems like BIOS V3.00 made these issues get worse? Also, did 3.00 allow you to run your memory at 3200-14? One more thing, when I try the OC Tweater method of Oc'ing the boot time to the first BIOS screen goes to ~20 seconds. When I use the Pstate method the boot time to the first BIOS screen goes back to ~10 seconds. Obviously the BIOS is going through some extra somethings to when using the OC Tweaker OC'ng method?!

Regarding the difference in using and not using the PStates (Custom = On, Auto = Off), the Ryzen power plan Minimum processor state setting, and the cores down-clocking, I agree that the cores do NOT down-clock unless the other, non-0 PState is set to Custom (in use.)

It's been a while since I have not had the PStates active, so I'll need to check that. But as I recall, I agree that the cores will not down-clock, even when using the Windows Balanced or Power Saver plan, with Minimum processor state set to 5%. That is, even with CnQ and C6 C State enabled. I do see the VCore dropping below 0.400V (in HWiNFO64) with the CPU power saving options enabled, and Minimum processor state set to 5%.

I have yet to find a way besides using the lower core speed PStates to get the cores to down-clock. That seems odd to me, since core down-clocking is not an exotic, new feature.

I'm also not sure about the length of POST time to the UEFI/BIOS screen when OC Tweaker over clocking, with PStates off. I'll need to check that. Do you have AM4 Advance Boot Training enabled or disabled? Currently the cold boot Windows Task Manager Last BIOS time today was 8.9 seconds.

My different board's different version of A-Tuning (ver:3.0.132) runs fine. Do you get that error when you run it?

Restart to UEFI with my board will not automatically put me into the UEFI IF my memory is over clocked. So it does skip going into the UEFI screen in that situation. I do not get any hanging of the system using Restart to UEFI, I'm not sure what you mean by that. I've mentioned the issue with Restart to UEFI in other threads, you must have missed it. No, you are not the only one to have that problem, although it seems you have more problems with it than I do.

Not much if anything you can do wrong using Restart to UEFI. It has worked flawlessly for me on... at least eight ASRock Intel chipset boards. The failure to work on my Ryzen PC is the first time I have seen it not work correctly, again ONLY when my memory is over clocked. Restart to UEFI is based upon a Window feature introduced in Windows 8, so I am unsure where the blame for its failure to work lies.

I'm still unable to run my memory above 2933 with AGESA 1.0.0.6a/UEFI 3.00. But I haven't spent much time trying. I have some new timings that supposedly work with Samsung B-die memory, different than the XMP profile values, but so far they don't make a difference.

Ken429 wrote: parsec, I agree, when I had the system down clocking using OC Tweaker the Vcore would drop to ~.400V which I found a bit troublesome since the Pstate2 said it should drop to .900V. Using the Pstate method of OC'ing the down clock does exactly what it supposed to do - 2200MHz and .900V (the values in Pstate2). I have AM4 Advanced Boot Training Disabled (the Default?). What is it and should I have it Enabled? When I say ~10 seconds to get to the first BIOS boot screen I mean from the time I hit the Power Button on the Case until the ASRock splash screen appears with the options to boot into the BIOS. Something changed with version 3.0 since now the Task Manager BIOS time changed from ~9 seconds to ~16 seconds. The 16 seconds is about the time it takes from the Power Button to Windows 10 appearing? I uninstalled A-Tuning and Restart to UEFI on the Ryzen system. When I said Restart to UEFI hung the system I meant it hung the system in the BIOS. The only way to get the system to work again was to turn the power off at the P/S turn it back on and Boot. I guess I'll keep my $200 dollars in the bank and keep using the EVGA 3000-15 memory. Please let me know if you ever get the GSkill Flare X memory running at 3200-14. I wonder if I should try the A-Tuning and Restart to UEFI programs for the SLI M/B download page since it appears to be newer versions? Can't be any worse than what I got from the Taichi M/B download page?!

Do you have Cool N Quiet and C6 C State enabled? It is the C6 C State that must cause the VCore to drop below 0.400V (mine goes to 0.384V). I've used my Ryzen PC configured like that for a long time, and I never have any problems. I also see a VCore of 0.900V at times, but the C6 C State must not be active at that time.

If I am not using the PState2 entry (all PStates set to Auto), with C6 enabled, I still see the VCore drop to ~0.400V, while the core speed is still at my OC of 3900MHz. The PC is stable at that time.

The AM4 Advance Boot Training option has been enabled by default in the past. I still think it is, but your board's UEFI defaults might be different. If you enable that option, the POST/startup time will be even longer, much longer. Are you sure it is disabled? What you describe about the Task Manager Last BIOS time sounds like AM4 Advance Boot Training is enabled.

Since Restart to UEFI does not work for me when my memory is over clocked, I rarely use it. I have never had Restart to UEFI hang the PC in the UEFI. The only thing it did does is, if I try using it, but then press the Del key when the PC restarts so I know I can get into the UEFI, I get into it fine. I can use it fine, but when I then click Save and Exit, the PC shuts off and stays off. I must press the power button to get it started.

So it seems Restart to UEFI behaves differently with your board or board's UEFI then my board.

Restart to UEFI is the same version on both of our board's download pages, ver:1.0.5. Restart to UEFI is a Windows program, and is the same version for Intel boards, ver:1.0.5 is the latest I can find. Again, it is based on a Windows feature, so I am not sure where the problem is, Windows, or Ryzen, or ??

A-Tuning on both of the X370 Killer SLI board download pages is version ver:3.0.132. The version on the X370 Taichi is ASRock A-Tuning Taichi utility ver:3.0.135, and dated a few days younger. So I don't understand why you think it is a newer version. Your board's Beta version is ASRock A-Tuning utility ver:3.0.131.

Are you using the ASRock A-Tuning Taichi utility ver:3.0.135 now, or what?