Temperatures, Noise, and Power Usage

Okay, as always most people will just look at the gaming numbers and a few CPU results and decide what they think. But it is always important to look at some of the other aspects when shopping for CPUs, GPUs, and of course barebones PCs like the NUC8i7HVK. This includes looking at overall power usage, temperatures, and noise. They normally all go together really. With this NUC it is especially interesting though because really this is a gaming PC with a normal integrated GPU on the Intel CPU as well as the Vega M GPU that is on the same package. CPU and GPU temperatures can transfer over to the other half and change performance.

To start things off I hooked up our Kill-A-Watt and dove into power testing. I went in knowing the included power supply is a 230 watt PSU. I was curious about CPU wattage so I opened up AIDA64 where the FPU test can put a huge and non-realistic load on the CPU. With just that running the system pulled 107 watts at peak. I then thought I could keep things simple and just turn on the GPU test along side of the FPU and Cache/Memory tests as well for good measure. Well with this setup it pulled 172 watts, but this is where things got interesting. Up until now really the testing had just focused on the CPU and the Vega M GPU. Well, I spotted something during this test. AIDA64 was actually seeing the included Intel iGPU as well and when it turns on the GPU test load it actually struggles to figure out what to do with it. So what happens is it switches back and forth, so both see 100% load then drop to nothing. In that situation, the NUC pulled 172 watts.

Now that’s a bit of a problem and it, as it turns out, would later cause issues in temperature testing as well. So I then jumped into 3DMark Fire Strike where I ran the combined test that does CPU for physics and the GPU. This only pulled 128 watts, but is actually I bet the best representation of actual in game power draw as the CPU was only reaching 24% load. So to push the CPU the rest of the way I turned the FPU test back on, finally, I saw both the CPU and Vega at 100% usage and the Intel iGPU wasn’t pulling a false load. I call it false because there aren’t any situations where the iGPU and Vega should be being used at the same time, so loading both wasn’t really giving us a real power draw.

power

So next I wanted to see just how warm the NUC was getting when under load. I could already really tell when it heated up in testing because it would go from quiet to loud. Here I just wanted to see how warm the CPU and GPU were getting. The CPU test was simple, I tested it using AIDA64 again. I tested at 100% fan speed and again with the stock fan profile. Remember this is a pain to switch between because you change these settings in the BIOS. I don’t want an award or anything, but a basic windows program to change fan profiles would be a nice Intel, just saying.

Anyhow at 100% fan speed the CPU reached 82c and was up to 86C without the cranked fan. Of course when under that hard of a load the fans spool most of the way up anyhow. Getting a GPU temperature result was a little more complicated. For starters, I normally use AIDA64 to load up the GPU and we know what happens there. But the big issue was figuring out what results were real. You have CPU package temperatures, a different temperature using the Radeon Wattman software, and then GPU-Z shows iGPU temperatures. They all heat up no matter if it is a GPU or CPU heating things up. I included all of my results to show that Wattman shows a much lower number when the rest of the CPU package heats up. I think the GPU numbers may be a little off. That said Wattman did show the GPU at 53c with the fans cranked and 63c on the stock fan profile.

temps

I didn’t stop there for thermal testing though. I busted out the thermal camera to get a look at how the NUC overall was handling things. I did this while loading up the CPU. The top doesn’t look too bad, you can see slightly warmer areas in the middle and along the right side. On the front, the hottest areas are the inside of the USB ports as they are near the CPU package on the motherboard.

thermal 1

thermal 2

Flipping the NUC over you can see a peak of the center where the CPU package is running warm but the two cooling fans on the bottom are doing a good job of pulling air over the heatsink that takes up the entire bottom of the NUC. The outside casing is reaching 90/100F. Then of course coming out of the back you can see where most of the hot air is being blown.

thermal 3

thermal 4

The last portion of my testing was just to take a look at the overall noise output of the two fans. To do this I tested at 100% fan speed to get an idea of how loud the NUC can reach. Then I went back and tested at idle and again under normal CPU load with the stock fan profile. The difference between all three results was staggering. At idle the NUC is basically inaudible. When I opened up some programs and got the CPU usage up in the 30/40% range you could start to hear the fan. This was the 39.9-decibel result. Audible but not a problem at all and for the most part, this is what you will experience. But then those two blower fans get cranked up god help us all, that is a loud 61.4 decibels. The two fans run at 3558 RPM, thankfully they are on the bottom. Flipping the NUC over with the fans cranked made it much louder.

noise

 

Log in to comment

We have 1839 guests and no members online

supportus