as soon as i start using QuickSync to encode something, Intel Desktop tools alert me that the VR (Voltage Regulator) temperature is critical.
I can run primt95 for hours without any such message but as soon as I start to encode something with QuickSync, the message appears.
At first I though that this must be just a bug (incorrect reading) but when viewed on IR (heat) you can clearly see that the visible VR parts (Box #2) heat up quite quickly, see: http://www.youtube.com/watch?v=FzlyCFFn9Wo&hd=1a
A regular view of the IR FOV: http://imgur.com/7nDAA
Is this normal or will this cause a problem?
Thanks for the help!
Which motherboard is it?
I've tried this on a DZ77BH-55K and when using QuickSync the reported VR temperature in Intel Desktop Utilities barely changes and stays around 38 degrees.
It is odd it ramps up so high on your board, although the IR image doesn't show it getting as hot as the reported temperature, which jumps up quite quickly. The only explanations I can think of:
1) The QuickSync function is drawing current from a different pin on the CPU than the main CPU cores and that pin on your motherboard is hard-wired to a single phase of the VR which is struggling on its own.
2) There is a problem with your CPU or motherboard
3) It's a reporting bug in the software or the sensor
thanks for your explanation/suggestions! It is the DQ77MK.
I see that your board has heatsinks on the VR, maybe that also makes a difference?
Yes, the temperature is not as hot as stated and the stated temperature jumps from normal (30 something) to 90+ within one second.
But i heard that if the temperature sensor is build into the component / part of the same semi conductor then the internal temperature might really be as high and jump as quickly and that the packaging might lag behind and also not heat up as much if it has a much bigger surface area.
In any case, I took a close up look at the VR components that are visible (2 of the (6?) phases in the top). What i noticed was that the chokes heat up more than the mosfets. But of course temperatures are again nowhere near stated. See: http://www.youtube.com/watch?v=oRcyeOYW3AY&hq=1
Also checked out the PWM controller but it also only hovers around 39C.
I guess I will try to make a simple app using the media sdk to repeatedly encode a sequence of randomly pregenerated frames, hoping the get the gpu usage as close as possible to 100%. Then i will just let that run for a week to see if something happens .
Forgot to mention, the QuickSync Load (GPU-Z GPU load percentage) varies greatly with the content of the video. If it is <80% the indicated temperatures will stay <75C. Only above 80% does the indicated temperature go above 90C.
i have the same problem. VR goes to 99 with WebGL or OpenGL 2.0 games like minecraft.
Test done with FurMark ( OpenGL Test )
Chrome WebGL Demos like this one
Running Win8 RPreview i7 3770 DQ77MK with Intel HD4000 Beta Windows 8 Drivers.
pls, Any idea to cool the VR? water, air, heatsink, change mbo?
I thought about ways to cool it too. However, since the temperature jumps to those above max temperatures very quickly and the external components that i observer do not show such a significant temperature increase, cooling the outside of the components would like not make much of a difference.
Maybe the packaging of the affected component is acting as an insulator and so that the component core has to heat up much more before sufficient heat transfer can be archived to keep the core at a constant temperature.
Due to my large CPU heatsink i was not able to observe all VR components as they are mostly arranged around the CPU socket. I also no not know which VR components are responsible for the integrated graphics voltage, if it the same as the CPU and over how many phases it is balanced. But since the components that i did observe did show a temperature increase (albeit not nearly as much as is shown) i do believe that i captured/viewed at least some of the phase components that are affected.
Another thing that i am wondering about is if other 77 series mainboards have the same internal temperature rise in the vr but that no one noticed because no tools are provided to read out the temperatures.
Next to the VR though, i also noticed that components on the left center (vertically) also visibly heat up quite a bit. I did not know what those components are but i did notice that the AMT features like KVM. IDE redirection and power control fail to respond after some time when the system has been running. Shutting it down for a few hours usually fixes this so i think that the heat up i see saw on the left side components is related to those failures.
I don't think it can be cooled, probably a component packaging problem. Attempts to cool these components may void the mainboard warranty. I will (haven't got around to it yet) stress test the mainboard for several weeks to see if I can get it to fail with the 'above maximum temperature'. If it does, i'll return the board.
Just a thought, are you using the retail fan and heat-sink? The retail ones are designed to move air over the VRs as well as the CPU, if you replace with a different one, like a vertical heat-sink, this movement of air over the VRs will not be happening, that might cause over-heating, there are warnings about this in the manuals:
A chassis with a maximum internal ambient temperature of 38 oC at the processor fan
inlet is a requirement. Use a processor heat sink that provides omni-directional airflow
to maintain required airflow across the processor voltage regulator area. If a non
omni-directional thermal solution is used customer might need to provide supplemental
cooling to the processor voltage regulator area.
As the components are quite small, they would heat up quickly given no air movement.
Hope that helps.
yes, thank you for pointing out the importance of airflow for cooling. Since I consider it sufficient in my setup i had not thought about mentioning it.
The problem with the stock heatsink and fan is that it is noisy to such a degree that I do not consider it usable. However, the case that I utilize uses forced airflow around the cpu and heatsink using 2 dedicated 120mm fans (so not counting fans for the rest of the case).
Since the temperature increase and corresponding warning is much quicker than the visible component temperature increase, increased air flow or other external cooling measures will likely not have any mentionable effect. But I will try it again with the stock HSF to see if the maximum temperature (reported and visible) is lower.
>>>But I will try it again with the stock HSF to see if the max temp is lower. reported and visible.
my HSF is the standard HSF with comes with the INTEL i7 3770. INTEL MBO with INTEL i7 with INTEL HSF.
I have the same Overheat on the VR with the INTEL HSF. Any OpenGL or WebGL Demo will overheat the VR within (Milli-) sec.
may be try temp. indicator like this one´s
or cooling spray with -45 grad C like this one
I will buy this things and will make some tests next week.
i will do some testing next week with cooling spray with -45 grad C cooling down the VR Chip and i will buy some temp. indicator sticker for max. 93 grad C temp. check with a response time within 1 sec.
this should show up in the INTEL VR Montior Temp. as fast as 1-2 sek.