Asus Z390 boards have a hidden +100mv offset

[dante_afk]

Just wondering why no big youtubers have acknowledged that yet, or in general no one is talking about it?

In general a Z390 board will take 100mv less for the same clockspeed as its Z370 counterpart, however the power draw and temperatures will be identical.

I have 3 examples (cba to scroll more back)

Example 1 - https://www.hardwareluxx.de/community/f139/oc-prozessoren-intel-sockel-1151-coffee-lake-laberthread-...
Example 2 - https://www.hardwareluxx.de/community/f139/oc-prozessoren-intel-sockel-1151-coffee-lake-laberthread-...
Example 3 - https://www.hardwareluxx.de/community/f139/oc-prozessoren-intel-sockel-1151-coffee-lake-laberthread-...

There are more examples if you scroll through the thread starting from 8th October onwards.

Just so you are aware, if you are running 1.3v on a Z390, you have in fact 1.4v etc.

[Shamino]

What you mean is software reads lower than what you are used to . It is routed to read directly from cpu die sense this gen , pls read the explanation of die sensing
https://www.overclock.net/forum/27686004-post2664.html?fbclid=IwAR28uwLxzdE6mG7Hya0w17pTYFoPwYTWNpuZ...

[mdzcpa]

I was getting the impression that v droop was not being controlled the same through LLC on the M11 series. Historically LLC 6 netted you about the same voltage readings (in CPUZ,AIS III, etc) as was set it the UEFI. More or less zero droop. M11 series requires 8. But what your saying is that previous voltage readings were just not as accurate and were overstated via more required inference.

This is good information.


And that said, LLC 6 still works just fine for me. I like a little droop for a 24/7 system.

[Shamino]

mdzcpa wrote:
I was getting the impression that v droop was not being controlled the same through LLC on the M11 series. Historically LLC 6 netted you about the same voltage readings (in CPUZ,AIS III, etc) as was set it the UEFI. More or less zero droop. M11 series requires 8. But what your saying is that previous voltage readings were just not as accurate and were overstated via more required inference.

This is good information.


And that said, LLC 6 still works just fine for me. I like a little droop for a 24/7 system.

yes, simply put check the graph by elmor, blue is the 'traditional' reading you are used to (also on other no Max boards),, orange is the current max v reading, green is actual cpu 'die' v,, the 12mv diff due to small diff in ground.

[Zammin]

dante`afk wrote:
Just wondering why no big youtubers have acknowledged that yet, or in general no one is talking about it?

In general a Z390 board will take 100mv less for the same clockspeed as its Z370 counterpart, however the power draw and temperatures will be identical.

I have 3 examples (cba to scroll more back)

Example 1 - https://www.hardwareluxx.de/community/f139/oc-prozessoren-intel-sockel-1151-coffee-lake-laberthread-...
Example 2 - https://www.hardwareluxx.de/community/f139/oc-prozessoren-intel-sockel-1151-coffee-lake-laberthread-...
Example 3 - https://www.hardwareluxx.de/community/f139/oc-prozessoren-intel-sockel-1151-coffee-lake-laberthread-...

There are more examples if you scroll through the thread starting from 8th October onwards.

Just so you are aware, if you are running 1.3v on a Z390, you have in fact 1.4v etc.

Wait, so the 9900k pulls 1.25V during AVX stress tests at stock speeds according to HWinfo64, but in reality it is pulling 1.35V?? that seems like a LOT of voltage for a CPU at default speeds..

One thing I've noticed is that different software reads the voltage differently. Under the conditions mentioned above (AVX stress) HWinfo64 reads 1.250V, while AIDA64 reads closer to 1.350V and ROG CPU-Z reads 1.144V... This is all very strange... 😕

[Silent_Scone]

Zammin wrote:
Wait, so the 9900k pulls 1.25V during AVX stress tests at stock speeds according to HWinfo64, but in reality it is pulling 1.35V?? that seems like a LOT of voltage for a CPU at default speeds..

One thing I've noticed is that different software reads the voltage differently. Under the conditions mentioned above (AVX stress) HWinfo64 reads 1.250V, while AIDA64 reads closer to 1.350V and ROG CPU-Z reads 1.144V... This is all very strange... 😕

The VID can vary a lot with recent Intel CPU. Regarding different software, you would need to contact the vendors, however, it is not recommend to run multiple polling applications at the same time as it can produce erroneous readings.

[Zammin]

Silent Scone@ASUS wrote:
The VID can vary a lot with recent Intel CPU. Regarding different software, you would need to contact the vendors, however, it is not recommend to run multiple polling applications at the same time as it can produce erroneous readings.

I'm talking about actual Vcore readings, not VID. and each application was run separately, not at the same time.

Can you advise an application that you know reads the correct Vcore on these motherboards and CPUs? Perhaps AI suite? I thought ROG CPU-Z would be accurate but it seems that isn't the case right now. Also, are you able to tell me what the Vcore should be for the 9900k on the ASUS Maximus boards under AVX loads?

For me I'm pretty sure it's pulling 1.250V on default settings, and with a large air cooler (Dark Rock Pro 4) the CPU hits 96C in RealBench stress test. It would be good to know the right monitoring software to use as well as the correct default voltage so I know my board isn't overcompensating the voltage to the CPU.

Thanks.

[HiVizMan]

There is no "one" or "correct" vcore for a range of CPU's simply because as you know each CPU is different.

All I can say is let temperature be your guide if your system does what you want and need it to do below an acceptable temperature then that temperature threshold is what your fine tune your voltages too. Choose any application and work exclusively with that application to fine tune the voltages, even if it is not 100% spot on it will close enough that it does not matter because you are using a different marker to set your limits namely temperature.

[Zammin]

HiVizMan wrote:
There is no "one" or "correct" vcore for a range of CPU's simply because as you know each CPU is different.

All I can say is let temperature be your guide if your system does what you want and need it to do below an acceptable temperature then that temperature threshold is what your fine tune your voltages too. Choose any application and work exclusively with that application to fine tune the voltages, even if it is not 100% spot on it will close enough that it does not matter because you are using a different marker to set your limits namely temperature.

So what you're saying is at default settings, multiple 9900k samples will have different load voltages under the same test conditions on the same motherboard? Forgive me if I've misunderstood.

My CPU is not yet overclocked, I can't do that until it's on water due to the excessively high temperatures on air (even though the cooler is huge) when using AVX stress loads. Once it's on water I'll start tweaking. I was hoping to determine what I should be seeing at default settings as I thought 1.250V and 96C seems excessive, but now I don't even know what software to use so that figure could be entirely wrong. And adding to the confusion, apparently the Vcore on Z390 is not the same as Vcore on Z370.. It's all very strange to me.