Noise Testing, SFF System Build, Temperatures
Sound and Noise Concerns – Those Damn Coils
When AMD launched the AMD Fury X card, one of the difficulties that AMD had was the annoying pump whine that permeated through the initial run of testing samples as well as the first batch of retail samples. AMD claims that a fix has been put in place for this but honestly we haven’t been able to get a hold of any other product from later production runs to validate that claim.
Because of that calamity I was disappointed to power on the R9 Nano for the first time on our test bed and hear the sound of coil whine from the card. This wasn’t the same problem as the Fury X, obviously, but instead the clear sound of inductors generating a buzzing sound that varies as the frame rate in the game changes. In a loading screen that is hitting 400 FPS? You are going to get a high pitched whine from the card. Playing a game at a modest 45 FPS? The sound will be a much gentler, lower pitched buzzing similar to that of a bumblebee.
Before you start ranting and raving about this, and trust me I did that initially as well, let’s talk through a few points. First, our test bed is an open-air environment that sits about 18 inches from my left ear. Any noise generated by the card is going to be easier for me to spot – and that’s on purpose. Second, when placed inside a case, the sound of the coil whine is muted pretty dramatically not just by the case itself, but from the sound of the air movement of the R9 Nano fan and the CPU fan.
The annoyance of this coil / PWM noise is going vary from user to user I’m sure, but the consensus in the office is that because the noise varies with the frame rate of the game it is MORE noticeable than the pump whine we found on the Fury X cards earlier in the summer. For the Fury X, there was no normal air movement from a fan on the card to help drone out the sound, but for the Nano, there is. Oddly, that is kind of an advantage in this scenario.
How did our normal sound level testing look?
Based solely on our sound meter, which only measures loudness and not pitch-specific intricacies, the R9 Nano and the R9 Fury X are nearly identical in terms of idle noise generation. The R9 Nano is a bit quieter under load, but again with a different kind of buzzing sound than the Fury X exhibits. The ASUS GTX 970 DC Mini card has essentially identical sound levels during usage but exhibits none of the coil whine that the AMD card has.
I wish I knew how to explain the coil whine away, but I just can’t. After the issues surrounding the Fury X launch I would have assumed that AMD would take the time to ensure there was no such issue with the R9 Nano. I was told that the R9 Nano is using the best quality inductors possible for this product, which tells me one of two things: either there are limitations on what kinds of inductors the R9 Nano can use or the compressed power delivery design on the 6-in PCB inherently causes noise issues that engineers haven’t been able to address. I’ll continue to poke and prod AMD for more specific information, but that’s the running theory I am working with here.
The Case for Nano – SFF Builds
I said this in the introduction to today’s review and in the R9 Nano preview story last month, this card is only targeting a specific demographic of PC enthusiast. You must want top level GPU performance, you must not be cost sensitive, you must want to run a very small form factor case and you must have a design that doesn’t allow for larger graphics cards. If you don’t meet those goals, the R9 Nano isn’t for you.
As an example of a system build using the R9 Nano that cannot be built to the same performance level with any other hardware, I present to you the PC Perspective "Nanosaurus".
This system is built around the Cooler Master Elite 110 Mini ITX chassis, capable of using a full size power supply but only permitting graphics cards with lengths as long as 8.3 inches. Technically this means the Fury X could fit in here though mashing in that water cooler is going to take some work. The reference design for the GTX 980 and GTX 980 Ti is 10.5 inches, much too large for this case. As are the current revisions of AMD’s own R9 Fury product. Obviously the new Radeon R9 Nano is the highest performing card you can fit inside this chassis and for NVIDIA that spot falls to the ASUS GTX 970 DC Mini, a card based on the GTX 970 with a length measuring 6.7 inches.
The MSI Z170I Gaming Pro AC Mini ITX motherboard – tiny Skylake platform!
The rest of this build is just as impressive as the GPUs though – we were able to get our hands on the brand new MSI Z170I Gaming Pro AC motherboard that supports the brand new Intel Skylake processors as well as USB 3.1 and 802.11ac wireless connectivity. That’s a lot of features and capability for a Mini ITX platform! In that CPU socket we placed the brand new Core i7-6700K and a Seasonic 650 watt power supply to keep it all running smoothly. Unfortunately for us, due to a mix up on the CPU cooler, we were forced to use an Intel stock cooler which was less efficient than I’d like and a little bit louder too. An Intel SSD 730 240 GB SSD held our Windows 10 installation and the games used for testing.
The result is an incredibly dense PC that sacrifices nothing in terms of performance and gaming capability. The Radeon R9 Nano provided significantly more GPU horsepower than the GTX 970 DC Mini could and keeping the card boxed up in the Elite 110 didn’t affect the gaming performance / clock speeds of the Fiji GPU in my testing.
Though the fan speed and temperature did increase with the move from an open-air test bench to the Cooler Master Elite 110 Mini ITX chassis, performance of the Radeon R9 Nano wasn’t affected in our testing.
Once installed in the case, combined with the sound of the R9 Nano fan and the CPU cooler, the coil whine we heard so blatantly in our open air test bed was a bit harder to find. Below you’ll see a YouTube video that demonstrates the sound differences as recorded by our Zoom H6 high quality audio device.
If you listen closely you can definitely hear the distinct buzzing sound of the R9 Nano in with the white noise of the case fan, CPU cooler and GPU fan. On the ASUS GTX 970 DC Mini configuration though there is no buzzing sound, only the white noise at a nearly identical level. Does this mean the coil whine is not an issue for the R9 Nano? Definitely not, in my opinion, it is still the biggest drawback to this product. This just demonstrates that depending on your installation configuration, room noise, levels, headphone / speaker usage, the coil whine may be less of an issue for some gamers than others.
I apologize for the lack of a graphic here, but I just remembered to add this data at launch this morning (thanks Sebastian!). On our open test bed, the Radeon R9 Nano never breached the 75C mark, which basically matches the results we saw in our set of R9 Fury reviews from both Sapphire and ASUS.
In our Cooler Master Elite 110 chassis, the temperatures got up to as high as 83C after extended gaming testing and the fans on both the R9 Nano and the GTX 970 DC Mini definitely got louder than they did on that same open test bed, but evenly so. You can get a sense of the noise levels of the combination of case fans, CPU cooler and GPU cooler in the video embedded above. Overall, I would say I was impressed with the cooler on the R9 Nano as it was able to keep the Fiji GPU in the same window as the much larger R9 Fury coolers we have seen at retail thus far.