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: 

DRAM Voltage Bug

JHughes
Level 7

‎10-07-2017 11:42 AM - last edited on ‎03-06-2024 08:51 PM by Community Admin

Hello, I have a Maximus VIII Gene with BIOS 3504, i7 6700k and 64GB of RAM.
I set my RAM to 1.35 Volt and it runs at 1.44 V instead.

My RAM, at rated speed, is specced for 1.35 volts. So I set the Ai Overclock Tuner to the XMP setting, which shows
XMP DDR4 3200 16-18-18-36 1.35V. This is correct per the label on the RAM.

Scrolling down to the DRAM Voltage setting shows that it has been set to the nearest voltage in BIOS which is 1.353. So far, so good.

I reboot to implement my new settings, and DRAM Voltage setting is still correctly set to 1.353. Excellent.
BUT, actual DRAM voltage shows 1.440, in BIOS and in Windows. Yikes, that's almost a .09V difference! 1.35 is not 1.44. 😞

After some testing I see that any voltage lower or higher than 1.35 tracks like it should. e.g. the next lowest voltage setting, 1.3464, produces 1.344, and the next highest, 1.3596 produces 1.360.

It's weird that the one and only one exact voltage specified by XMP is bugged. I don't know that 1.44 volts will fry your RAM but I'd hate to unknowingly have it happen. And 1.35 is an extremely common RAM voltage, esp. for people using ROG motherboards.

I've never seen this before, does anyone have an idea? Is it only with the M8G mobo?

64GB RAM 2X Corsair LPX 32GB (2x16GB) 3200MHz C16 DDR4 DRAM Memory Kit, Black (CMK32GX4M2B3200C16)
https://www.amazon.com/Corsair-2x16GB-3200MHz-Memory-CMK32GX4M2B3200C16/dp/B016ORTNI2
Labels:
0 Kudos
17,640 Views
10 REPLIES 10

z1n0x
Level 7

‎10-11-2017 12:14 AM

Hi JHughes, i'm in the same situation as you.

Setting XMP crank up the voltages

DRAM Voltage is 1.440 while 1.3530 is actually set per XMP profile
VCCIO Voltage went from 1.160V to 1.336V
System Agent Voltage went from 1.144V to 1.320V
Standby Voltage went from 1.168V to 1.343V

I noticed from the MEI thread that you've updated to 11.7.4.3314 aswell.

From what i can tell it appears to be issue with Intel MEI firmware 11.7.4.3314 and Bios 3504, whether from bug in the MEI firmware or incompatibility between the both, who knows.
With the previous firmware version 11.7.0.3290 and Bios 3504 voltages were in the normal.

I guess it's what we get for applying those unofficial updates in desire to be on all the latest stuff, thinking somehow it will works better, with fixes for bugs we don't experience or performance improvements we won't notice. :rolleyes:

I don't know where else to report that issue, so Intel is notified about it and hopefully fix it.

For now i'll turn off XMP, to get the voltages back to normal.
0 Kudos

z1n0x
Level 7
In response to z1n0x

‎10-12-2017 12:05 AM

I noticed something about the voltages under XMP and MEI 11.7.4.3314 / Bios 3504.

So, my Kingston HX430C15SBK2/16 has 2 XMP profiles.

- 2666 14-14-14-36 1.35V
- 3000 15-16-16-39 1.35V

Under 2666 voltages are as they always were.

DRAM Voltage - 1.3530V
VCCIO Voltage - 1.160V
System Agent Voltage - 1.144V
Standby Voltage - 1.168V

Under 3000 however, voltages go ballistic

DRAM Voltage - 1.440
VCCIO Voltage - 1.336V
System Agent Voltage - 1.320V
Standby Voltage - 1.343V

Prior to MEI firmware 11.7.4.3314, XMP 3000 used to have the 2666 voltages.

Whether, it's MEI frimware bug, incompatibility with Skylake's 3504, some issue with XMP3000+, or specific Maximus VIII Gene, i don't know.

I had this machine for 20 months now, i've updated the MEI firmware multiple times, never had an issue until now.
0 Kudos

bochi_mode
Level 7
In response to z1n0x

‎05-06-2018 12:37 PM

Hej guys

I have the same problem, too. I use my one xmp profil with a corsair vengeance lpx 2x8gb 3200 cl16 rams and the bios writes that it set to 1.3530 (with yellow) but at little left is 1.4400 ?(its the real voltage). Without xmp it is 1.2.
Now what I can do, I use the ai suit 3 and there I made a profil with 1.3530. Now Its fine but every time (after boot) I have to set my profil in Windows to use this normal xmp voltage. By aida64 now its 1.344.
Under the bios I don't know how can I solve it. I want to use the xmp of course.

The bigger problem that I'm useing this config. 1,5 years ago. And just now I realised that my rams run on 1.444 V. However I don't know which bios update after started this high value.

"Asus please do something!"

i7 6700k
maximus viii gene (latest bios)
lpx ddr4 3200 cl16
0 Kudos

Korth
Level 14

‎10-12-2017 01:54 AM

1.44V is unlikely to damage your RAM, unless it's making it run too hot. Hot RAM can be cooled with better chassis airflow or a dedicated RAM cooler.

JEDEC-2133, JEDEC-2400, and JEDEC-2666 DDR4 standards specify DDR4 voltages from 1.20V to 1.35V.
DDR4 silicons are designed to comply with JEDEC. DDR4 memory controllers (like those built into CPUs) are designed to comply with JEDEC (or to advance JEDEC, lol, since Intel is such a mighty member of the group).

But Intel XMP implicitly permits frequency overclocking and increased voltages, even though Intel's technical literature does not explicitly define upper XMP voltage parameters. Several sources claim that Intel has "unofficially" stated maximum DDR4 XMP voltage is 1.50V.
I've seen DDR4 kits rated 1.55V, 1.6V, and 1.65V (at least in the early days of DDR4, circa 2014-2015). Some people report pushing their DDR4 as high as 1.85V (as if it were primitive DDR3!) - while using memory waterblocks, often also using motherboard volt-mods - but let reckless fools be reckless fools if they stubbornly insist on breaking their own hardware, lol.

I've never pushed the upper limits, but I believe I read somewhere on these forums that ASUS BIOS settings won't accept values which exceed preset "safe" thresholds. I might be wrong about this.

The values marked on your RAM parts (and even your motherboard) are really more marketing specs than technical specs. The most extreme DDR4-4266+ kit available today is still built from (binned) DDR4-2666 silicon and, technically, your processor guarantees nothing over its rated (DDR4-2666) performance because XMP is a form of overclocking. (Although it's still fairly realistic to expect stable 2933MTps~3200MTps performances with high-quality ASUS motherboards and high-bin DDR4 kits.)
"All opinions are not equal. Some are a very great deal more robust, sophisticated and well supported in logic and argument than others." - Douglas Adams

[/Korth]
0 Kudos

JHughes
Level 7
In response to Korth

‎11-11-2017 12:54 PM

Thanks Korth. BTW, what would one use to check RAM temperature? One of those handheld infrared thermometers?

z1n0x wrote:
Under 3000 however, voltages go ballistic

DRAM Voltage - 1.440
VCCIO Voltage - 1.336V
System Agent Voltage - 1.320V
Standby Voltage - 1.343V

Prior to MEI firmware 11.7.4.3314, XMP 3000 used to have the 2666 voltages.

Whether, it's MEI frimware bug, incompatibility with Skylake's 3504, some issue with XMP3000+, or specific Maximus VIII Gene, i don't know.

I had this machine for 20 months now, i've updated the MEI firmware multiple times, never had an issue until now.

z1n0x is it coincidental that both of us have the Maximus VIII Gene? And I'm afraid the latest MEI update hasn't helped.
https://rog.asus.com/forum/showthread.php?91661-Intel-MEI-updates&p=685524&viewfull=1#post685524
0 Kudos

Korth
Level 14
In response to JHughes

‎05-06-2018 05:04 PM

JHughes wrote:
Thanks Korth. BTW, what would one use to check RAM temperature? One of those handheld infrared thermometers?

You could use a non-contact IR thermometer. Not the most accurate instrument for every possible application, some shiny IR-reflective stuff on your parts or IR-opaque glass barriers or incorrect emissivity will skew all your temp measures.

You could use thermal probes - plugged into the motherboard or into any compatible instrument like a DMM.

I often use a cheap indoor/outdoor "household" thermometer for this sort of stuff - it's accurate enough (±0.5°C/1°F or so) for general electronic work and it has a handful of handy wired/wireless remote temp probes which let me monitor/log real-time temps "in situ" without leaving the chassis open.

And we're just talking about RAM. It doesn't (shouldn't) get terribly hot anyhow. If it does then put a heatspreader on it and/or fan airflow across it.
"All opinions are not equal. Some are a very great deal more robust, sophisticated and well supported in logic and argument than others." - Douglas Adams

[/Korth]
0 Kudos

MMikhail
Level 9

‎11-12-2017 04:43 AM

Here I have no problems with voltages and ram is at good temps

68882
0 Kudos

JHughes
Level 7
In response to MMikhail

‎11-12-2017 07:09 AM

Have you selected the XMP profile in BIOS?
0 Kudos

JohnnyTimeBomb
Level 7
In response to JHughes

‎11-12-2017 07:29 AM

I had no Ram voltage issues with this build until I just updated like everyone else. Now it is almost at 1.5v I guess I'll manually bring it down some that seems too high. I ran aida64 and there were no ram speed changes with the update or voltage changes.

before update after update
ram 1.344v 1.488v
vccio 1.272v 1.208v
agent 1.272v 1.208v
Asus maximus hero ix
Intel 700k 5.1ghz
Evga 1080 ti sc2 hybrid 2054mhz
Corsair lppx 32gb (4x8) 3200mhz
Samsung evo 1tb nvme
Corsair H115i
Thermal Grizzly Conductonaut on cpu and gpu
Noctua industrial fans all locations radiators included
Phanteks evolv atx tempered glass version
0 Kudos