Hero VIII DRAM voltage discrepancy

So, no love finding an answer to this via Google or the ASUS forums.

I just recently swapped out my Z170x-Gaming 5 for a Maximus VIII Hero and I'm fiddling with the tuning of my memory and cpu settings. I've noticed something that's bugging the crap out of me, and possibly affecting the tune, but I can't seem to resolve it.

My DRAM voltage shows up as 1.36v, whether it's set to 1.36v or 1.365v. If I bump it up to 1.37v, then it jumps to 1.376v. It will allow the setting in the bios, but ALL monitoring software, including the BIOS itself after a reboot (MC Fast boot IS disabled) will show the DRAM voltage being set to either 1.36v or 1.376v, nothing in between. It does not fluctuate.

I've tried increasing the DRAM current setting to allow up to 110%. 120%. Default value of 100%. No effect.

LLC is set to 3, which allows the CPU to be stable, and the CPU is currently overclocked to 4.6Ghz with 1.33v core voltage. Everything is stable, including my tightened timings of 13,15,15,28 (XMP is 15,16,16,35), but it bugs the crap out of me that when set to 1.365v, that is not what is reflected in HWinfo, BIOS, or any other monitoring utility.

Any idea? Is there a setting I'm missing on Maximus VIII boards that resolve this discrepancy?
 
I don't think so. The board documentation and everything I see on the Maximus VII series guides specifies DRAM voltage steps of .005v, and I don't think that would be specified if those increments were verboten.

My Gaming 5 board, and a good many other boards from Haswell on that I've worked on have had various different possible increments ranging from .005v to only allowing .020v steps like my Gaming 5 was, which is part of the reason why I changed boards, but all of them always reflected exactly what was set in the DRAM voltage setting of the bios, which in this case is 1.365v. I don't believe a setting of 1.365v, when allowed by the DRAM voltage setting, should be reflected in a changed value of a full increment lower or higher than that setting when viewed in the sensor data in windows.

I could be wrong, but I'd like to see verification that this is normal rather than just having to go with it without actually knowing. Seems pointless to have the capability of using .005v increments in the settings if that's not what's going to be used.
 

TJ Hooker

Titan
Ambassador

Where did you find that information? I just skimmed through the user manual from the Asus product page, didn't see any mention of step size.
 
You're right, it's not in the manual. I think it was in the ASUS official Maximus VIII series overclocking guide. I assure you that the increments are .005v though. They set in .005v increments, and they stick, they just do not reflect the actual settings when you check the sensor values in Windows.

I know voltages can fluctuate, but these don't. They stick firmly at that voltage, which is of course NOT the voltage they are supposed to be at. After doing a little research it looks like there have been DRAM voltage bugs throughout the entire Maximus VIII series of boards on more than one bios version, but not on some other versions. I may have to try an older bios image and see if that solves the issue.

Another thing I've noticed is that my fan operation is not normal regardless of whether I configure for PWM or DC operation in the bios. Fans are all Noctua NF-A14 PWM fans except the second fan on the backside of my heatsink (NH-U14S with dual fans) which is a Noctua NF-A14 industrialPPC-2000 PWM model.

With a setting of (Any given value, say, 65°C), set to configure 100% PWM operation at that temperature, or even when set to DC mode, the fans will initially respond and max out at their normal maximum RPM, but then they'll gradually ramp down from around 1500rpm to around 1000-1100rpm, even though the CPU core/package temp is still above the temperature configured for 100% fan speed operation. So, for example, when I fire up Prime Small FFT or Blend, temps of course jump up above my reference point, which by now I've set at 60°C trying to get them to remain at full fan speed, but even so after a minute or so they'll drop in RPM down to around 1000rpm, bouncing between there and 1200rpm invariably, rather than remaining maxed out at between 1450 and 1600rpm which is where they should be at full RPM.

Doesn't matter what I set the full speed temp to or whether it is configured for PWM or DC, and all fans are set to be controlled by CPU temp, and all fans do this, not just the CPU HSF fans.

I'm beginning to wonder if this board doesn't have problems and needs to be returned to ASUS.
 

larrycumming

Prominent
Aug 15, 2018
422
0
410
sounds like a bios bug

well, a normal system ram voltage is supposed to fluctuate in hwinfo because voltages are "monitored and regulated" by the pwm/vrm in a feedback circuit. asus went with a digital design so it has more freedom of control thru uefi/bios. a simple bios update may just fix it and should be first to try
 
An update isn't going to be much help, since this is the latest bios. I'm thinking perhaps a rollback to an earlier bios version but haven't decided yet. I might just live with the discrepancy for now and see if they release any further bios images for this board. I'm also thinking that after I see what the 9th gen reviews look like I might just sell this whole system off and upgrade to a Z390 with an i7 that's got soldered TIM anyhow.
 

larrycumming

Prominent
Aug 15, 2018
422
0
410


Yea something isn't right about that board from your description.

alternatively you should be able to send it in under warranty, they will be able to do a full reflash of bios. depending on your locale, the service center may even do a full reflash free of charge or for a minor fee for out of warranty boards.

the 9th gen definitely looks like a good choice however.
 


Overclock

CPU Core/Cache Voltage @ 0.005V increment
CPU Graphics Voltage @ 0.005V increment
CPU VCCIO Voltage @ 0.0125V increment
CPU System Agent Voltage @ 0.0125V increment
DRAM Voltage @ 152-step
PCH Core Voltage @ 88-step
BCLK/PCIE freq. 40MHz ~ 500MHz @ 0.01MHz increment

https://www.custompcguide.net/the-differences-between-all-asus-mainstream-z170-motherboards/

152 step works out to .005v increment, exactly as the CPU core and graphics voltage listed above. I also found the specification in ASUS staffer Raja's overclocking tutorial on the Z170 ROG boards.
 

TJ Hooker

Titan
Ambassador

How did you translate # of steps into step size?
 
You are aware that in the bios, where a specific number or option is not REQUIRED to be chosen, hitting the plus or minus sign will increase or decrease the value by one increment, yes? In this case, on this board, that increment is .005v for the DRAM voltage field.

The maximum DRAM voltage setting allowed on this board is 2v. The stock JEDEC SPD dram voltage for DDR4 2133mhz is almost universally 1.2-1.225v depending on the module and motherboard. The difference between 1.225v and 2v is .755v.

.005 x 155= .755. It is close enough to infer based on what the incremental changes to the voltage are when increasing or decreasing using the + and - keys. Also, this is what is referenced in practically all overclocking documention regarding memory on ASUS Z170, Z270 and Z370 motherboards. Seems plain that it stands to reason.