Lower RAM voltage to decrease CPU temperature

[Makentox]

Asus z170-z i7-6700k @ 4.7 1.320v. DDR4 2133MHZ
Hello, i want to decrease cpu temperature by lowering ram voltage. Stock RAM voltage is 1.2 v i have tried to lower it down to 1.150v and it decreased my temperature from 82/84/84/76 package= 83 down to 77/82/78/75 package 78. So around 5C down just by tweaking RAM voltage. This will give me a temperature headroom to achieve 4.8GHZ on my cpu. My goal is 85C max during ITB stress test.
So the question is how can i check if my voltage is stable? So far i run memtest86 and passed with 0 errors. Will memtest86 give me an error if my voltage too low?
Some people say its bad to run RAM at low voltages,is it? is 1.150 considered as low?
And finally is it good idea or bad? If intel XTU or HWmonitor are accurate in reading temperatures then 5C decrease in temperature, isnt it great?
Hopefully you can understand what this thread all about, thanks!

EDIT: Updated screenshots and settings in BIOS to manual voltage for current testing.
Core/cache 1.300
RAM 1.150/1.250
VCCIO 1.000
Agent 1.100
PCH 1.100
CPU SB 1.000





HWmonitor above 1.15V RAM voltage, 1.25V bellow RAM voltage, other settings remain the same, Fan speeds are on max during testing. Also want to point you that not all cores get effected the same way, looks like core 0 and 2 has the most impact from RAM voltage!

 

[Lochekey]

What is the exact voltage you are tweaking? Normally adjusting ram voltage will not effect cpu temp.

 

[Makentox]

What is the exact voltage you are tweaking? Normally adjusting ram voltage will not effect cpu temp.

Well DIMM voltage. I came up with this idea when i was overclocking RAM, when i increased voltage my cpu temps went up. This why i tried to lower voltage to see if it helps. I have no idea why it works this way but this what hwmonitor and intel XTU report. Also as u can see it cant be heat from RAM sticks that heatsync absorbs because mostly core 1 and 3 has decreased temperatures.
Or else what i can think of 1. once i lower RAM feed voltage it automatically lowers something else at the same time in bios settings 2. Bugs temperature monitoring software. Just what it could be? My cpu remains stable no matter what RAM voltage is.

 

[Lochekey]

Other than the DRAM voltage did you adjust any other voltages like VCCIO or System agent Voltage or are these on auto?

 

[Makentox]

Other than the DRAM voltage did you adjust any other voltages like VCCIO or System agent Voltage or are these on auto?

No i havent... If i go any moment now back to bios and only change RAM voltage, then stress test with ITB cpu temperature will vary depending if i increase or decrease RAM voltage. This why i made this thread because i couldnt find any similar post.
Also VCCIO and system agent voltage are on AUTO, i will go to bios right now and check if values will change after i decrease or increase voltage. Btw before further discuss answer my questions from OP.

 

[Lochekey]

I like to use HwInfo 64 to monitor temps and voltages. Unless you get out a dmm though and actually check the leads you will have to trust the software.

I'm not real sure on memtest as I do not use it myself.

As for the temperature drop 5c is pretty decent.

 

[Makentox]

Well HWmonitor never lied temperature before, and i sure that intel software would not lie too. Ill edit OP with some picture to prove it.
Maybe anyone else can check it out and reply, will be much appreciated.

 

[Lochekey]

Im sorry but it is hard for me to accept that the fluctuation of DRAM voltage will directly effect core temp. I have tried DRAM voltage from 1.1 to 1.6 volts in increments of 0.1v on my 6700k and the core temp shows no difference when stressed with IBT. I suspect something else is going on with the system or the motherboard is adjusting some other voltage to see a temperature variation that high.

By the way, why do the 2 HwMonitor screenshots have different available values to be displayed?

 

[Makentox]

Hmm which exactly values do u mean? Well mistake is taht i didnt make a screen shot while under test and made pics just after i stopped stress test. But if u mean that the first has more voltage values than other is just HWmonitor, sometimes it shows sometimes it doesnt. Well i suspected the same, must be ram voltage automatically changes something else in bios, logically. Something that has nothing to do with cpu voltage or else it wouldn't be stable. Also the weird thing is that it mostly has influence on core 1 and 3. Cant explain it at all Could it be memory controller inside CPU chip that produce more heat when voltage increased? I have never experienced it myself as well.

 

[Lochekey]

Yea i was talking about the extra voltage values. It seems like a flaky program if it is doing that. It could be the IMC voltage adjusting causing the difference in temperature. You could try manually setting your VCCIO and System agent voltage then try testing again. If it stops showing this behavior I would suspect one of these values is auto adjusting on you. I would also try a different monitoring program to see if it is displaying similar results.

 

[Woomack]

Max temps are 82 vs 84 and lowest are 75 vs 76. It's pretty normal and I don't think it's affected by DIMM voltage. Memory is separated and its voltage is not affecting anything else. It even has its own power section.
Memory controller ( SA ) and IO voltages are affecting temps. Also PLL and some other voltages are affecting CPU temps.
If raising VDIMM is causing higher temps then motherboard is automatically adjusting other voltages ( hidden under auto settings ). It's happening on some motherboards but I've never seen higher IO/SA voltages when you set higher VDIMM.

 

[EarthDog]

Max temps are 82 vs 84 and lowest are 75 vs 76. It's pretty normal and I don't think it's affected by DIMM voltage. Memory is separated and its voltage is not affecting anything else. It even has its own power section.
Memory controller ( SA ) and IO voltages are affecting temps. Also PLL and some other voltages are affecting CPU temps.
If raising VDIMM is causing higher temps then motherboard is automatically adjusting other voltages ( hidden under auto settings ). It's happening on some motherboards but I've never seen higher IO/SA voltages when you set higher VDIMM.

Exactly. DIMM voltage shouldn't have anything to do with your core temps unless other voltage values, Woomack listed, are changing as well.

 

[Makentox]

- - - Updated - - -

Max temps are 82 vs 84 and lowest are 75 vs 76. It's pretty normal and I don't think it's affected by DIMM voltage. Memory is separated and its voltage is not affecting anything else. It even has its own power section.
Memory controller ( SA ) and IO voltages are affecting temps. Also PLL and some other voltages are affecting CPU temps.
If raising VDIMM is causing higher temps then motherboard is automatically adjusting other voltages ( hidden under auto settings ). It's happening on some motherboards but I've never seen higher IO/SA voltages when you set higher VDIMM.

Thanks for your respond, i have updated my screenshots and set most settings in bios to manual fixed voltage, please look OP.
Please answer my questions from OP about stability memory testing.

 

[Makentox]

Yea i was talking about the extra voltage values. It seems like a flaky program if it is doing that. It could be the IMC voltage adjusting causing the difference in temperature. You could try manually setting your VCCIO and System agent voltage then try testing again. If it stops showing this behavior I would suspect one of these values is auto adjusting on you. I would also try a different monitoring program to see if it is displaying similar results.

I have tried different monitoring programs which never lied to me before, example aida64, hwminitor, intel XTU (which is official intel software...)
I set voltages u asked to manual value, please check OP

- - - Updated - - -

Exactly. DIMM voltage shouldn't have anything to do with your core temps unless other voltage values, Woomack listed, are changing as well.

Hi could u try`to answer questions in OP?

 

[EarthDog]

You can keep going down this path but...

To answer in order...
1. Yes, it should throw an error if the sticks are not stable.
2. No. It does not hurt the ram to run lower voltage. It can become unstable though.
3. 5C drop is great, but, we don't see how dimm voltage would do that explained above. It's like you are saying adjusting air conditioning makes the engine run cooler. Two unrelated things.

 

[Makentox]

You can keep going down this path but...

To answer in order...
1. Yes, it should throw an error if the sticks are not stable.
2. No. It does not hurt the ram to run lower voltage. It can become unstable though.
3. 5C drop is great, but, we don't see how dimm voltage would do that explained above. It's like you are saying adjusting air conditioning makes the engine run cooler. Two unrelated things.

Thanks, i have never done any ram test before because ram never gave me much in game performance, never worth the time making it stable. So good to know that if my voltage is too low i will get an error during memtest86 test.

 

[Tír na nÓg]

Voltages, except from vCore being on auto, it is likely that some secondary voltages lik Sys Agent or I/O vary according to the DDR settings.

 

[Makentox]

Voltages, except from vCore being on auto, it is likely that some secondary voltages lik Sys Agent or I/O vary according to the DDR settings.

Read posts, i said i set voltages manually at least those that are possible to control. But still what you say is correct and this what we already understood, please answer OP questions if u dont mind.

 

[Tír na nÓg]

I am with Earthdog on this one: 2 unrelated things that should not have any impact on each other.

Did you redo the operation a few times? To make sure it is not a matter of ambient temp, or just a couple of random temp peaks that occured?

 

[Rainwater]

So are each of your individual Core Temps all 5c lower or just your Package Temp?