• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

VRM overheating, throttling i7 ivybridge

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
4+2 phases are more than enough and shouldn't throttle under load at reasonable voltages. I have 4+2 phases on my current Z370 mobo and it runs 6 cores at 5GHz+ without issues.
Throttling is generally because of too high temps or too low power limit. Overheating VRM are not causing throttling, they are causing instability.
If you turn off all C states and other power saving options and set power limit higher ( not sure if your board has it but on some can set current limit to some impossible values ) then it shouldn't throttle but will shutdown once CPU is too hot and looking at some posts here, it's not overheating.
 
It's expanded and I still don't see it in XTU. Downloaded hwinfo64 and it does show a tag for package/ring thermal throttling. That tag does not flag as positive under p95 load when the cpu drops back to 35x.
 
I don't know if they changed it on recent XTU, but current throttling didn't use to be a default item and you have to configure it to show it. I don't like using XTU to monitor as it is rather CPU hungry while it is running taking several %.
 
It is default. I didn't change a thing. You will need to resize the window/panel in XTU where you see all of the readings... see image below. See the red line? You should be able to drag that line up and make it bigger thus showing the rest of the relevant options.

As far as it taking up CPU cycles, it does that, yes. On my system its using .1-.3% on idle (16c/16t) it will occasionally blip up, but isn't constant. It isn't much (on my system) is the take away here. Chrome with 6 tabs is using 1% for the record. But since we are running stress tests and looking for throttling, none of that really matters.

xtu.jpg
 
Last edited:
Did that, it's just empty space w/o corresponding checkboxes to add it in config (specifically motherboard vr thermal throttle, but there are others to add)

I did notice that 'turbo boost power time window' for my rig is only listed at 1 sec (if memory serves, away from the rig now) vs. 16s on yours.

I will also dig back into bios and make sure that all the speedstep , c6 and related settings are disabled.
 
Odd. I have the latest version, and when I pressed reset to defaults, nothing changed. I wonder if Ivybridge is supported properly in XTU.
 
In BIOS can you increase power limits and max current to see if it will prevent throttling? With VRM throttling the Clock speed should go down and up cycling forever like this Video. When running mosfets to TJmax there should be no stability problems.

 
Last edited:
My limits are already set to maximum, 500 watts for this board.

Here's XTU from my system.
 

Attachments

  • Screenshot (1).png
    Screenshot (1).png
    95.1 KB · Views: 126
  • Screenshot (2).png
    Screenshot (2).png
    102.5 KB · Views: 126
Last edited:
Does not look like XTU shows VRM thermal throttling for your motherboard. Did you try HWiNFO to see if there is thermal throttling?
 
wingman99 said:
did you try HWiNFO to see if there is thermal throttling?
Click to expand...
Yes, no

The old laser shows vrm sink surface temps in the high 40C range, chokes up to 60C. More work to do.
Comments on 'turbo boost power time window' setting?
 
Using infrared thermometer my VRM heat sink reads 60c with the VRM MOSFET Tj 100c reading in hardware HWiNFO, No throttling.
 
Last edited:
wingman99 said:
In BIOS can you increase power limits and max current to see if it will prevent throttling? With VRM throttling the Clock speed should go down and up cycling forever like this Video. When running mosfets to TJmax there should be no stability problems.

Click to expand...


So this is what happens when overclocking goes mainstream? You have to machine your own sinks because the ones they made are complete ****? All show and no go? Back in the day, I left AMD because their motherboards kept crapping out. Never had a problem since, and now this is what you get for your hard earned money? That's almost enough to make me want to try out AMD again. Lol.


Edit:

I haven't seen any kind of throttling like that, that's new to me.. But I think I may have turned that stuff off. Ive had this cpu at 100c, maybe a bit more before she thermaled.
 
That video was a bit dramatic. I've reviewed around 12 X299 motherboards and none showed VRM throttling with a 7900X at 4.5Ghz on an open air test bench (Running latest P95). Wait, maybe a Mini ITX board did? I don't recall off-hand.
 
EarthDog said:
That video was a bit dramatic. I've reviewed around 12 X299 motherboards and none showed VRM throttling with a 7900X at 4.5Ghz on an open air test bench (Running latest P95). Wait, maybe a Mini ITX board did? I don't recall off-hand.
Click to expand...

If I recall correctly you did not use prime95 small FFT like the reviews I have seen.


LINK:
 
Well.. That's kind of one of the reasons why I went back to air cooling, from using an H100. I used to have to get creative with fan placement, now.. not so much. Big cooler, big fans, move air. So much easier. But now I wonder just how hard are my vrm's working.. When I was able to feel the vrm sinks as long as I had a fan blowing on it they stayed cool. Even with no fan they didn't really feel remotely hot, luke warm maybe..
 
wingman99 said:
If I recall correctly you did not use prime95 small FFT like the reviews I have seen.
Click to expand...

Indeed. For power testing, it is P95 Blend. The VRM/OC Testing is done using OCCT for 30 mins. I should have said OCCT out of the gate.

Part of the issue is auto voltages as well going wonky during testing. Setting a manual voltage keeps things under control as well.

I still feel this was blown WAY out of proportion though considering its really one of the only things that does it....and not if you are careful with your overclock. ;)
 
Last edited:
good stuff there. I tried occt and it takes slightly longer to throttle, perhaps 5 min instead of 3, and not as persistently.

The kicker is that there's probably some headroom in this chip even with my aircooler. I'll have to sort out the vrm cooling first, apparently.
 
Some data points:
With the rear 120mm case fan detached and blowing at the VRMs at an angle, throttling would improve but not disappear. Since the board was pushing Vcore of 1.33 under load at 'auto' for 44x, I went down to a negative offset such that peak draw is closer to 1.3v, reducing throttling further but not eliminating it.
At this point I should probably drop the offset as far as I can to be stable at 44x and call it a day.

Going the other way, up, it looks like I'm hitting a voltage wall. 46x after upping offset to get 1.4v at load still crashed small FFT fairly quickly. I was hoping for 47x, which may need ~1.45v but probably not with this board.

Or I can get creative with a dremel/hacksaw and fab bigger sinks for the vrm's from some old socket A parts.
 
You must log in or register to reply here.
Back