Help
cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 

Above X52 multiplier breaks certain auto voltages

Apothysis
Level 7

‎12-02-2019 04:31 AM - last edited on ‎03-06-2024 07:21 PM by Community Admin

Hi! I'm on the Asus XI Apex and I noticed something really odd and potentially dangerous whilst I was trying to dial in my 5.3 GHz overclock. Once you set an all core multiplier, CPU Standby Voltage and CPU PLL OC Voltage are set to very high values without any other change. Reading voltages in BIOS/HWInfo64.

To clarify, the scenario I had before:
Core multiplier: 52
CPU Standby Voltage: 1.07v (Auto)
CPU PLL OC Voltage: 1.2v (auto)

After performing only one change to BIOS settings, increasing core multiplier:
Core multiplier: 53
CPU Standby Voltage: 1.6v (Auto)
CPU PLL OC Voltage: 2.6v (Auto)

This seems like an absolutely crazy change for the bios to do on its own. I fixed the issue my manually setting Standby Voltage to 1.05v and PLL Bandwidth to 0 (which brought it back to 1.2v). So far these are the only two voltages I could find that are affected but there's a ton of voltage settings that just say "auto" with no revealed auto values. Before noticing these absurd voltages I was getting CLOCK_WATCHDOG_TIMEOUT BSODS which I've never seen before while overclocking a 9900KS. I'm still stress-testing but I suspect its related. Has anyone run into similar issues?
Labels:
0 Kudos
1,278 Views
2 REPLIES 2

Falkentyne
Level 12

‎12-03-2019 02:05 PM

Apothysis wrote:
Hi! I'm on the Asus XI Apex and I noticed something really odd and potentially dangerous whilst I was trying to dial in my 5.3 GHz overclock. Once you set an all core multiplier, CPU Standby Voltage and CPU PLL OC Voltage are set to very high values without any other change. Reading voltages in BIOS/HWInfo64.

To clarify, the scenario I had before:
Core multiplier: 52
CPU Standby Voltage: 1.07v (Auto)
CPU PLL OC Voltage: 1.2v (auto)

After performing only one change to BIOS settings, increasing core multiplier:
Core multiplier: 53
CPU Standby Voltage: 1.6v (Auto)
CPU PLL OC Voltage: 2.6v (Auto)

This seems like an absolutely crazy change for the bios to do on its own. I fixed the issue my manually setting Standby Voltage to 1.05v and PLL Bandwidth to 0 (which brought it back to 1.2v). So far these are the only two voltages I could find that are affected but there's a ton of voltage settings that just say "auto" with no revealed auto values. Before noticing these absurd voltages I was getting CLOCK_WATCHDOG_TIMEOUT BSODS which I've never seen before while overclocking a 9900KS. I'm still stress-testing but I suspect its related. Has anyone run into similar issues?


There's no "CPU PLL OC Voltage" setting in the BIOS?

You get clock watchdog timeout (clock signals just freeze) when PLL OC voltage is too low. You can trigger that by setting a PLL OC voltage of 1.05v and trying to run any stress test. (in fact values below 1.15v can start acting weird). But I don't know how you enter values directly. The Gigabyte bios lets you enter the voltages.

Can PLL OC voltage even go up to 2.6v? I could have sworn the maximum value was 2.1v for LN2 ....(its for some cold bug stuff).
0 Kudos

Apothysis
Level 7
In response to Falkentyne

‎12-03-2019 03:57 PM

Falkentyne wrote:
There's no "CPU PLL OC Voltage" setting in the BIOS?

You get clock watchdog timeout (clock signals just freeze) when PLL OC voltage is too low. You can trigger that by setting a PLL OC voltage of 1.05v and trying to run any stress test. (in fact values below 1.15v can start acting weird). But I don't know how you enter values directly. The Gigabyte bios lets you enter the voltages.

Can PLL OC voltage even go up to 2.6v? I could have sworn the maximum value was 2.1v for LN2 ....(its for some cold bug stuff).


Like I mentioned it's slightly disguised as "PLL Bandwidth", adjusting this to 0 sets the PLL OC Voltage back to the default 1.2v. The UEFI states that a value of 6-8 is recommended for high frequency overclocking, but I found that a value is 6 is what it also applies at auto with a multiplier of 53. Indeed, it goes even higher than 2.6v..

Interesting. Is it possible that a frequency of 5.3 GHz might also need a slight bump in PLL OC Voltage from the default 1.2v? I've been getting weird clock watchdog bsods ~40-60 minutes into Realbench with no L0 errors or anything like that. I've never encountered this bsod due to low vcore at lower multipliers so it seemed weird that I'd suddenly get it now.
0 Kudos