ThrottleStop undervolt - higher VRM temps?

Someone on the channel made this comment in my recent video

"DO NOT use CPU UV 24/7, it is not good for the motherboard. CPU runs cooler yes, but it stresses the VRM on the motherboard more and if the VRM cooling is not good enough you will get microstuttering because the VRMs are overheating and in the long term your mb may die."

We had a brief discussion and I began to research this. I found a few posts out there that have asked about this same thing. Not much was resolved from these topics and research seems limited to its defense.

What are your thoughts?

If you would like to see the conversation and video that started this topic, it's here

 

by that logic all notebooks would die and having your PC on idle kills it.

 

That wouldn't make much sense to me - reducing voltage usually results in lower thermals. What the VRMs do is take the +12V (or in this case I believe it's +15V) from the power supply and tone it down. Reducing the voltage applied to the CPU also reduces the required power since, IIRC, P=IxU. Sure, the VRMs aren't 100% efficient but their efficiency doesn't really depend on the CPU VID, does it?

 

Exactly

 

I think his main line of thinking is: lowering the voltage will cause the amperage to increase (which is even more closely related to heat than voltage).

However, the supply voltage going to the VRMs themselves remains the same coming from the battery/mobo regardless of what output you are seeking. When we select that we want to undervolt the CPU more than stock, we are telling the VRMs they have to buck even more voltage (to ground?) than they were originally. I assume this causes additional heat and stress on the VRMs themselves but it's hard to say how much exactly.

Does anyone know if the VRMs outputting the lower voltages are designed to maintain the same wattage output (this would cause amperage to go up, right?) Or if by lowering the voltage coming out of the VRMs the total overall power output in watts gets reduced and the current adheres to a preset curve?

I think the main point to consider is whether operating your system at a lower total temperature across the board is more beneficial for the longevity of your system as a whole or if slightly reduced stress*** on a single component with higher overall system temps full time is the better way to go.

I think lower temps across the board is the better way myself but that's just my $.02. I would love to hear from some electrical engineer types.

Btw, is it the case that when undervolting, a CPU is more likely to run into current throttling issues due to needing a higher amperage limit in order to maintain the same maximum power output?

 

Fundamental law: When the voltage decreases and resistance stays the same, the amperage MUST also decrease.

That's why a small undervolt can go a long way - reducing the voltage by 10% will also reduce the amperage required by, say, 10%. Thus, a 10% undervolt would result in a 19% (1 - 0.9 x 0.9) power consumption drop. (Voltage x Amperage = Wattage)

If anything, undervolting should be better for the VRMs, since they no longer have to maintain the same wattage. Undervolting should not cause the amperage to increase... Modern laptops will throttle its VRMs if they overheat, but the worst killer is probably voltage spikes from the AC/DC charger (storms, etc).

Yes, buck converters (Which are the majority of VRMs on laptops) are more efficient when the voltage they are dropping is lower (19v dropped to 15v is more efficient than 19v dropped to 5v) but most CPUs can undervolt by 0.2v at the most, so VRMs may be a little less efficient, but the drop in current and voltage requirements from the CPU negates any inefficiency.

logically, CPUs run on CV (Constant Voltage), meaning the VRM choose the voltage (requested by the CPU, of course) and the CPU draws whatever the current it needs. If a motherboard is current limiting, the motherboard sends a signal to the CPU to reduce its clock speed (and thus voltage, followed by current) so that the CPU throttles to the point where it no longer crosses the current limit

 

Hello everyone,

I made a comment regarding this in Bob's video and he addressed me to this thread so i'm gonna try to explain the more detailed that my level of English allows me to what happened to my Laptop.

First of all, my Laptop is a Lenovo Legion Y720. I did a repaisting with Arctic Cooling MX-4 plus undervolt of -0.120V only to the CPU using XTU and a Powershell script which ran with a Scheduled Task each Windows Login.

Temps were always very good for an i7 7700HQ, never passed the 80° mark, I think the cooling system of this Laptop is OK.

So I'll try to explain the issue, about a month ago while I was playing I began to experience video and audio stuttering (I don't know if this is the right term... something like a very aggressive audio and video lag), so I closed the game and started to troubleshoot the issue to try and find what was causing it.

Temperatures were OK for all the hardware. Taskmgr was OK, CPU / GPUs / Disk / Network / RAM, everything 0% but for example when opening Chrome the CPU would go from 0% to 100% for about 5 seconds. My first guess was that the issue was related to Nvidia Optimus so I downloaded DDU and went to safe mode to perform a clean installation of the Intel and Nvidia drivers (now this stuttering wasn't only while gaming, I could barely navigate and took a very long time to download DDU... Mouse was also laggy and input keys were laggy too). Now when I logged in to safe mode I realized that this issue was bad, because it persisted even in safe mode... anyways a proceeded with the clean reinstall of vide drivers which didn't solve the issue. So I turned off the laptop, went to bed and to work the next day. When I came back from work the issue was miraculously solved like WTF. After a month of everything working excellent last Thursday this issue started again so this time I began with hardware testing. Memtest was OK. Disk was OK so yeah I kinda suspected from the beginning but didn't want it to be true that this was a motherboard issue anyways I remembered I don't know why about the undervolt and about the scheduled task so I put everything back to 0 and disabled the task, restarted the laptop and everything started to work just fine again after a few windows restarts that issue re-appeared so I opened the laptop and removed the SSD I was using (1Tb Samsung EVO 850) and installed the HDD Seagate that came with the laptop. With the Seagate Windows wouldn't even start so I tried resetting Windows and it started still with the same stuttering but now also I tried to for example connect my external monitor or even connect a mouse or external keyboard Windows would hang and Windows would not even start with a peripheral connected. So now the Lenovo is in the warranty.

Ok, I'm sorry for the very long post but any guesses are very much appreciated. I don't know if this issue can be related to undervolt or just I got a faulty laptop. Maybe all the Legion Y720 shipped to Argentina are all faulty or maybe not.

 

This can be caused by multiple things really, undervolting is great, but only to a certain degree, don't forget that you undervolt an adaptive voltage, which will cause fluctuations on the CPU which sometimes will underpower it causing those microstutters you're talking about. Personally I have throttlestop running 24/7 and have 2 modes, one for battery mode where I keep adaptive voltage running as usual and limting my turbo speeds and one where I have a static voltage for performance, this makes my CPU run very stable on very low voltage without crashing on me- I suggest you try the same as well. Also look for possible Bios and EC updates which might fix this very issue, since this might be EC related as well.

You best bet is to look in your event viewer to see what kind of errors you're getting.

 

I think you're facing an issue of bad mobo. I suppose @Freitz had this issue where he got random WHEA errors in HWINFO and in event viewer when undervolting and on stock.
For the record, MX4 isn't even a good paste and has less longevity.
Try cleaning the gold contacts of RAM, SSD and HDD using a clean cloth or even a microfiber coth and keep them aside for 2-3mins and then put them back to see if the issue is fixed. I had a problem where my ssds were not detecting suddenly and random stuttering as well.

 

Thank you Gintoki! BIOS was up to date last time I checked, I updated it when I changed the HDD for the SSD and performed a clean Windows install. I didn't had a chance to check Event Viewer because it's last state was the BIOS check and it would hang right there so I was forced to send it to the warranty.

I'm changing this Lenovo for an Asus GL503VS SCAR... I don't know if i'm going to undervolt it.

 

I know MX4 is not the best but is what we can get in Argentina sadly... MX4 and Silver 5.

 

Gelid or ICD or even Cooler master should be available in most countries. Just check this thread out http://forum.notebookreview.com/thr...nd-apply-traditional-and-liquid-metal.806840/
Did you try my suggestions?

 

I didn't had the chance to try that because the laptop is being reviewed by the warranty service right now... 2 passes of memtest showed 0 errors and this issue was consistent with 2 different disks so my first thought was that the motherboard was dying (considering that if I plugged a peripheral like the monitor or a mouse or a keyboard the system would just hang and would not load with these peripherals plugged).

What would be your top 5 for thermal pastes, i'm gonna check to see if I can get one of those instead of MX4... I thought MX4 was doing a very good job on my laptop with stock paste the CPU would reach about 85° and after repaisting it could hardly reach 80°.

 

Personally I am using CM Maker gel nano which is quite big and its good. I could afford only this since Gelid, Thermal grizzly and other liquid metal were very expensive in India.

 

Thank you kind sir. Look at the bright side, at least you can buy Thermal Grizzly in your country

 

It was priced 5 times than Cooler master paste which I mentioned.

 

Last update: Confirmed by Lenovo warranty service, it was a faulty motherboard.

Can someone explain how can I check the VRM temperature and if it can be checked with HWMonitor and if possible normal VRM temperatures in idle and under load. Thanks!

 

If Lenovo actually allows sensors to query VRM temps you can use HWINFO and in preferences, enable PCH device and then restart HWINFO and select/click on sensors and scroll down to EC where you get CPU,GPU, VRM etc.. temps and fan speeds(if supported)