Our latest quiet PC build guide is optimized for 4K gaming and features an Intel Core i7-6700K processor, an Asus GTX 980 Ti Strix OC graphics card, and a Corsair Carbide 600Q tower case. Can we achieve silent (sub-15 dBA@1m by our definition) operation?
February 2, 2016 by Lawrence Lee
At SPCR we use a variety of displays for day-to-day operations but given the nature of our work, none of them are particularly high resolution. Recently we acquired a 4K TV, strictly for a side project, and it presented an interesting opportunity: The chance to build and test a quiet 4K gaming build. A 4K (aka UHD) display has a native resolution of 3840×2160, quadruple that of 1080p, for a ridiculous total of 8.3 million pixels. Spread across a large display, this arguably offers a more immersive gaming experience than a triple-monitor configuration with bezels in between, and a more usable workspace for non-gaming activities. It’s not a cheap endeavor however as even some high-end GPUs can’t deliver suitable frame rates. Given this kind of horsepower, component selection is critical to limit the noise output on such a PC.
GPU: Asus GTX 980 Ti Strix OC – US$700
The star of any gaming PC is the video card, and this is especially true if
you need to drive a 4K display. The sheer number of pixels that need to be rendered
demands a monstrously powerful GPU with a substantial amount of video memory.
The Asus GTX 980 Ti Strix OC is easily the most expensive component in our build,
but for a top tier GPU, it’s actually quite competitive. The GTX 980 Ti performs
substantially better than the GTX 980 and AMD’s current flagship, the R9 Fury
X, and is similar to that of the much more expensive Titan X, whose only advantage
is an insane 12GB of video memory that most games can’t utilize. The Asus Strix
OC model is one of the fastest variants and we’ll need every bit of speed we
can get. Usually, you can get better value by selecting a more affordable graphics
card, but in this case, the minimum requirements are just too high. A mid-range
card would slow to a crawl with the latest and greatest titles even at the lowest
detail levels, so you really have to reach for the best.
Even with a formidable GPU, maxing out the settings in most games won’t result in a smooth experience as outputting four times the resolution of 1080p is simply too demanding. The extra pixels are still worth it if you have to turn the image quality settings down a few notches. If that’s not acceptable, then you’ll have to look into much hotter and louder multi-GPU configurations. The GTX 900 series is incredibly energy efficient but the 980 Ti is the exception given its 250W TDP. To deal with this kind of load, the Strix OC edition is equipped with a substantial heatsink underneath a trio of fans but even so, it’s not nearly as quiet as the various GTX 980’s and 970’s on the market. Multiple 980 Ti’s would require some serious aftermarket cooling to produce some semblance of quiet. Our build allows for this possibility if you so desire.
Case: Corsair Carbide 600Q – US$150
Many cases marketed as quiet follow a faulty paradigm, sealing up the enclosure to prevent noise from escaping and restricting airflow in doing so. Such designs are likely quieter than the average chassis assuming the components inside generate the same noise level. In reality, the lack of proper ventilation means higher fan speeds are required to produce an equivalent level of cooling. Often this extra noise more than offsets the any potential savings. It sounds counterintuitive but airflow is key for a quiet case.
An unusually tall and wide case, the Corsair Carbide 600Q is not a particularly attractive tower (the windowed 600C is preferable if aesthetics are important) but it has a wonderfully open airflow design conducive to cool and quiet operation. The intake vents along the sides of the front bezel are generously sized, there are no drive cages or other obstructions blocking the front fans, and the dust filters are pleasantly unrestricted. It features an inverted motherboard orientation, 5.25-inch drive support in case you still rely on optical media, noise damping material, and a triple fan speed controller. It doesn’t offer much in the way of mass storage though, with only a pair of 3.5-inch drive bays.
If you’ve blown your budget on the graphics card, the more affordable Fractal
Define S has a similar layout that works almost as well, and the Define R5 is
similar but more versatile as it has removable hard drive cages. The SilverStone
RV05/FT05 should also be considered as they are the best performing sub-$200
case we’ve ested due to their rotated motherboard layout and immense stock fans.
CPU: Intel Core i7-6700K – US$370
Aside from price and the sheer number of cores offered (which is not of great significance when it comes to gaming), Intel has had a massive edge over AMD when it comes to high-end desktop CPUs for the better part of the last decade. Intel chips are superior in IPC (instructions per clock) and consume significantly less power and therefore generate less heat, something that is of utmost importance for a quiet PC. This may change when AMD drops their much anticipated new Zen CPU architecture later this year but until then, Intel is remains firmly planted in its throne.
There are three platforms to choose from: the older Haswell (LGA1150), the
latest Skylake (LGA1151), and the extreme enthusiast Haswell-E, soon to be Broadwell-E
(LGA2011-v3), with the latter giving users access to powerful six/eight core
chips. Modern gaming titles are unlikely to be bottlenecked by any decently
clocked quad core model, so it doesn’t really matter which socket is chosen.
We selected Skylake simply because it’s the most modern/future-proof but the
flagship i7-6700K could be replaced with a slower model without any detriment.
Skylake doesn’t offer a huge performance upgrade over Haswell, so the older
LGA1150 socket is certainly viable if you want to stretch your dollar a bit
but the price difference isn’t that big. Haswell-E’s ultra expensive hex/octa-core
CPUs are likely to be overkill and might be considered only if you’re simultaneously
live streaming video to Twitch while gaming or something similar.
CPU Cooler: Scythe Mugen MAX
We’ve used the Scythe Mugen Max in previous builds and it has yet to let us
down. It’s an excellent cooler with a superb mounting system, a pleasant sounding
stock fan, and its asymmetrical design creates plenty of clearance for the DIMM
slots. PC builders often select a large CPU heatsink and memory with tall heatspreaders
only to discover later that they interfere with one another.
Editor’s Note: Scythe’s Mugen 4 and Mugen Max were two Scythe
products that needed a design fix for the odd users with a Skylake mounting/cracking
problem (thinner substrate and too long screws led to the cooler breaking
the CPU with case movement). They probably shortened the screws by a turn.
Closed loop liquid coolers are incredibly popular for custom gaming PCs but they’re quite expensive and the noise from the pump is almost always awful, ruining the idle noise level. They’re also unnecessary as the CPU doesn’t get worked very hard by most games. It makes more sense to use watercooling on video cards as they typically have much higher power requirements than their CPU counterparts and in most enclosures, GPU cooling is always a challenge due to the relatively centralized position of the video card far away from most fan placements.
Motherboard: Asus Z170-A – US$150
The Z170-A is not a particularly glamorous or otherwise noteworthy motherboard,
but like most Asus models, it’s reliable and offers excellent fan control. It
has seven fan headers with all but one individually controllable with customized
control in the BIOS as well as through their superb Fan Xpert utility. We only
need four fan headers but it certainly doesn’t hurt to have more. A keystone
feature of the Z170 chipset is overclocking via multiplier and this is complemented
by the Z170-A’s digital 8+2 phase power regulation system and decently sized
VRM heatsinks. Rounding out the features are SATA Express and M.2 storage options,
as well as USB 3.1 with both Type-A and Type-C connectors at the back. The plastic
shroud over the I/O ports is a bit of an eyesore and may inhibit the ability
of the VRM heatsink to dissipate heat so it was removed for this build.
Many users overpay for expensive models with bells and whistles but many of the extras just don’t benefit the purpose of this system — you could game just as well on a B150 or H170 chipset board. The main drawback of going this route is that due to PCI-E lane limitations, a proper dual GPU configuration with x8/x8 bandwidth would be ruled out, restricting you to single GPU upgrades down the road. That’s one reason you might select a Z170 model to pair with an standard CPU with no overclocking capabilities.
Power Supply: SilverStone Strider Platinum 750W – US$140
From a noise perspective, the power supply was one of the most complained about
components when PC silencing first started to gain traction. It’s one of the
reasons that the PSU is now found at the bottom of most modern tower cases.
Back in the old days, the PSU was always situated at the top, getting pounded
by the heat coming off the processor, and thus requiring a much higher internal
fan speed. Power supplies also weren’t very efficient but today, every major
power supply manufacturer produces models that are compliant with the 80 PLUS
standard. With less heat waste being generated, PSUs are naturally quieter,
and this is especially true of 80 Plus Gold and Platinum units.
The SilverStone Strider Platinum 750W is a fully modular power supply with
a 120 mm fan that shuts off when under 20% load. For a single video card configuration
it’s a bit overkill as a 600W rated unit would fulfill power requirements with
room to spare. However, there is merit to trading up as efficiency peaks in
the middle of most PSU’s ranges and their fans are usually tuned with this in
mind. It also allows provides enough headroom for a second GPU to be added down
Corsair and Seasonic make a variety of excellent quiet alternatives and if you have the cash, the Corsair AXi, HXi, and RMi series are of particular interest as they have the ability to manually control the fan via an internal USB header.
SSD: Samsung 850 EVO 500GB – US$150
PC enthusiasts today have more storage interfaces to choose from than ever before but U.2 and SATA Express drives are still in their infancy, and while M.2 offers more potential bandwidth for solid state drives, in real life operation, the difference is minor compared to SATA 6 Gbps, so you won’t miss out by going with the more traditional standard. As for capacity, solid state prices have come down enough that it’s feasible for a pure gaming rig to avoid mechanical drives altogether.
Samsung makes some of the best SSDs on the market and their current generation
mainstream consumer 850 EVO is competitively priced at both the 500GB and 1TB
levels. If a hard drive or two is required for storing massive amounts of media
or the like, a 5400 RPM WD Blue/Red drive is ideal as they are extremely quiet
and produce little vibration.
RAM: Kingston HyperX Fury 16GB (2x8GB) DDR4-2666 – US$110
On Skylake systems, memory speed and timings have very little effect on gaming performance, and as RAM is essentially a commodity, it’s perfectly reasonable to simply select the cheapest RAM from a brand you trust. For our system we’re utilizing a pair of Kingston HyperX Fury 8GB DDR4-2666 sticks (actually part of a quad-channel kit). 8GB is sufficient for any single game (in fact many games are 32-bit and can’t utilize more than 2GB of system memory) but with memory prices being so low, it would almost be foolish not to move up to 16GB.
FANS: Phanteks PH-F140HP x 2 – US$35 & Scythe GlideStream 120 – US$10
The 600Q is an excellent case but its not with its faults. The included 140 mm fans cool just fine but emit a rough unpleasant sound unfitting a truly quiet PC. The Phanteks PH-F140HP on the otherhand has an incredibly smooth acoustic profile and just happens to be one of the best best performing 140 mm case fans we’ve ever tested. It also comes in a few different blade/impeller colors if you care about color coordination. Note: the similar looking square-framed F140SP and F140XP do not sound nearly as good as the F140HP.
Another issue with the 600Q is that it is not designed for 140 mm fans with
120 mm mounting holes and while we managed to get two of the Phanteks fans installed
at the front, there simply wasn’t any room at the back. The rear exhaust fan
was swapped for a Scythe GlideStream 120 to complement the Mugen MAX’s GlideStream
140. The performance difference between 120 and 140 mm fans is negligible and
while the GlideStream doesn’t sound quite as smooth as the Phanteks, it’s good
enough, and positioned at the rear of the case, furthest away from our ears.
DISPLAY: Vizio M43-C1 43-inch 4K Smart TV –
Normally the display in these build guides is not discussed, but for 4K it’s
worth mentioning. Quality 4K monitors are egregiously expensive and text size/scaling
can be an issue on the most affordable (smaller) panels, making a 4K TV an attractive
alternative. The Vizio M series is widely regarded as one of the best budget
4K TVs on the market. While we don’t have much experience with such displays,
we have to agree with this consensus. The 43-inch M43-C1 is a beauty with fairly
solid color reproduction, excellent contrast, and a fantastic price-tag. The
most important factor for 4K PC gaming is input lag, which the M43-C1 addresses
with a low lag gaming mode that didn’t produce any noticeable ghosting. It also
seems fine for regular computing, e.g. browsing, reading, but I haven’t spent
more than 15 minutes at time using it in such a capacity. From a hardware perspective,
its main drawback is only one of its five HDMI ports is HDMI 2.0/HDCP 2.2 compliant,
making it the only port that can output 4K at 60 Hz.
Assembling a system inside the 600Q is straightforward, though experienced
builders may feel some unease due to the inverted motherboard orientation which
positions many of the connectors upside-down relative to the norm. With plenty
of clearance between the front/bottom of the chassis and the motherboard edges,
the interior is relatively roomy. The fitting of the fan mounts was the only
issue encountered during the process.
The 600Q’s fan placements are designed for both 120 mm and 140 mm fans but the designers assumed the mounting holes would be in the corners. Like many 140 mm fans, our Phanteks PH-F140HPs have 120 mm mounting holes, so the alignment was slightly off at the front due to the larger frames. As a result, I had to use zip-ties to partially secure the top intake fan.
A similar issue can be found at the rear. The 120 mm mounting holes are offset rather than centered, so physically, a 140 mm fan using those holes is impeded by the lip of the chassis. Thus, the use of a Scythe GlideStream 120 mm fan instead of a third Phanteks. The CPU heatsink is equipped with a 140 mm GlideStream so there is some symmetry to this choice.
Looking inside the completed build, it’s clear why the 600Q performs well. The intake fans are quite close to the video card with no obstructions in the way. The floor offers additional airflow as it’s fully ventilated and nicely elevated.
The three case fans were connected to the motherboard for full customizable control so the 600Q’s fan controller was tied up and hidden away with the rest of the cabling. There are a variety of strategically placed cable tie-down points but the 2.5-inch trays can be used to route wires as well.
With plenty of room behind the motherboard, I didn’t feel the need to pin every wire down flat. The side panel installed without difficulty even with noticeable cable slack.
Measurement and Analysis Tools
When the system was initially set up, the size of the text and icons was ridiculously
tiny, so increasing the scale to 150% in Windows 7 was the first on my to do
list. Later I found that some games would not allow me to select the native
resolution, with the maximum option being 2560×1440. This was addressed by entering
the Nvidia Control Panel and manually setting the resolution and refresh rate.
My favorite aspect of Asus motherboards is the Fan Xpert module found inside their AI Suite utility. After calibrating the fans to determine their operational range, the fan speed curve for every controllable fan can be altered with four configuration plot points. Spin up and spin down times are adjustable and you can designate which onboard temperature sensor is linked to each fan. Unfortunately, it’s not smart enough to pick up the GPU temperature sensor but this is a feature they’re working on (for Asus graphics cards only) for the future. You can even rename the fan headers and designate where in the case they’re located so you don’t mix them up. Temperature and fan speeds are reported at the bottom of the interface with an option to swap the plain text readings for a charting function.
For controlling the GPU, the included GPU Tweak II utility is fairly competent.
Both the main tuning interface and chart are customizable so you can hide settings/sensors
you don’t use. The graphing feature is a bit limited as it can only display
three attributes without scrolling. The previous version of GPU Tweak is more
capable in this regard and is probably preferable, at least for testing. The
user defined fan speed curve has a high level of granularity with up to nine
adjustable plot points. Like our power supply, the GPU fans are turned off and
silent under light loads. By default, the fans don’t spin up until the GPU hits
Before any stress testing let’s take a quick look at the noise produced by each part. This was done by leaving the system on but idle and stopping/unplugging all the fans to isolate individual components. Our ultra-quiet hemi-anechoic chamber has a noise floor of 10~11 dBA@1m and
the baseline of this system measures barely above that.
SPCR 4K Gaming PC: Component Noise Levels (idle)
Exhaust Fan (Scythe GlideStream 120)
Intake Fan (Phanteks PH-F140HP)
Measuring mic positioned 1m at diagonal angle
Ambient noise level: 10~11 dBA@1m.
The idea behind the above testing centers around the fact that the GPU fans
will be doing most of the work in this system as the GTX 980 Ti is both more
power hungry and will be stressed more than the i7-6700K. After establishing
the approximate the GPU fan speed necessary, the remaining four fan speeds were
tweaked to complement the noise level being generated as not to drown out the
GPU fans. For example, if we’re happy with the GPU fans running at 1600 RPM
(20 dBA), it would be impractical to operate the exhaust fan at 1200 RPM (25~26
dBA) as it has far less impact on temperatures and be needlessly noisy.
As it turned out, the CPU and exhaust fans were practically identical noise
generators when set to similar fan speeds, so they could be adjusted in unison
to the same speed. The larger Phanteks fans at the front were a bit louder,
so they were set to slightly slower speeds to compensate.
For real game stress testing we used the Resident Evil 6 Demo Benchmark Tool (which is handy as it loops automatically after a brief pause at the end) and the built-in benchmark of Total War: Attila. A combination of Prime95 and FurMark was also used to stress the system beyond realistic norms.
Brief session of testing at 1080p with the case fans running at the minimum controllable speeds revealed that under normal gaming conditions, the GPU fans would run anywhere between 1200 and 1900 RPM under stock fan control while the GPU clock rarely dipped below 1418 MHz. I found that the remaining fans running at 650~700 RPM (not much higher than the minimum allowable) while also reducing the GPU fan speed manually (just enough not to lower the clock speed) generated the lowest overall noise output. Switching over to 4K proved to be a tad more demanding with the system drawing 5~6W more power from the wall and the GPU clock lowering slightly from 1418 MHz to 1405 MHz, even when the GPU fans were left at their default settings.
System Measurements (4K)
Resident Evil 6 (Demo Benchmark, peak)
Total War: Attila (In-Game Benchmark, peak)
Prime95x2 + FurMark
GPU Fan Speed
System Power (AC)
* GPU fan speed manually set.
CPU and rear exhaust fan at 700 RPM, intake fans at 650 RPM (19 dBA@1m at idle).
Ambient temperature: 21°C.
The card abides by its default target temperature of 83°C stringently, keeping the GPU core at around 80°C though I found that the GPU fan speed could be lowered to produce quieter results without reducing the clock speed and therefore performance. This raised the GPU temperature by 5°C but this seems to be acceptable. The clock speed drops a bit more when it hits 87°C; If these temperatures were dangerous, it would throttle much harder.
Of the two game tests, Resident Evil 6 was harder on the system, requiring a GPU fan speed of 1920 RPM (50%) at its peak, which brought the overall noise level to a reasonable 26 dBA@1m while a much quieter 23 dBA@1m was possible with a slight GPU fan speed tweak. Most games are not particularly demanding on the CPU, so the CPU temperature remained comfortably under 50°C.
Average Framerate: 1080p vs. 4K
Resident Evil 6
Total War: Attila
70.9 fps (quality)
58.9 fps (max)
50.3 fps (quality)
27.2 fps (max)
While Resident Evil 6 is physically taxing on the GPU, it’s not very demanding. It produced a score rather than framerates but watching it in action, it stayed above 120 fps most of the time at 1080p, often eclipsing 200 fps. This was with all the quality settings maxed out by default. Moving to 4K, the score dropped by more than half, and averaged 55~60 fps which is quite playable.
Total War: Attila on the otherhand requires a lot of horsepower, staying under a 60 fps average at 1080p using the “maximum” graphics preset. Oddly, this is one notch below the “extreme” preset which actually renders the most eye candy. The middle-of-the-road “quality” preset allows it to run somewhat smoother though it dropped below 30 fps here and there during the test. At 4K, the maximum setting was only smooth during short stretches so it needed be turned down at least one level. That being said, it still looked absolutely gorgeous as the additional pixels had a far greater effect than all the various tweakable quality settings.
I found playing at 1080p resolution on the 4K screen to be a very enjoyable
experience. It’s not as crisp as proper UHD, but the Vizio M43-C1 does a surprisingly
good job at upscaling the image. If you’re thinking about upgrading to 4K for
productivity purposes but don’t have the hardware to play games properly at
the higher resolution, with the right display, you can game comfortably at the
lower resolution rather than shelling for an expensive graphics card.
Running at minimum fan speeds, our system registered just 14 dBA@1m sitting
idle, a level that was barely audible inside our anechoic chamber. In a normal
environment, you wouldn’t be able to hear it operating except at very close
proximity. At the most stressful point of the Resident Evil 6 test, our system
measured a respectable 23 dBA@1m, with the GPU fan slightly lowered (without
affecting the GPU clock speed) compared to stock and the remaining fans running
at 700/650 RPM. The type of noise being output was also quite innocuous thanks
to the acoustically unobtrusive fans selected for this build.
The GPU fans were the most noticeable as they were a higher pitched than the
others but they didn’t stand out too much. The GPU also suffered from coil whine
but this was only annoying when looking at menus and loading screens rather
than during gameplay, and only when V-Sync was disabled. The much higher framerates
on these screens caused it to squeal so capping it at 60 fps with V-Sync helped
quite a bit. I also found found that the background sound/music of Total War:
Attila set to a moderate volume was enough to mask the noise of the PC altogether.
These recordings were made with a high resolution, lab quality, digital recording
system inside SPCR’s own 11 dBA ambient anechoic chamber, then converted to
LAME 128kbps encoded MP3s. We’ve listened long and hard to ensure there is no
audible degradation from the original WAV files to these MP3s. They represent
a quick snapshot of what we heard during the review.
Each recording starts with ambient noise, then 5~10 second segments of product
at various states. For the most realistic results,
set the volume so that the starting ambient level is just barely audible, then
don’t change the volume setting again while comparing all the sound files.
SPCR’s Quiet 4K Gaming PC Component List
Street Price (USD)
Intel Core i7-6700K
Intel Core i5-6500
Scythe Mugen MAX
Coolermaster Hyper 212 Evo
Kingston HyperX Fury 2x8GB DDR4-2666
G.Skill Ripjaws V 2x4GB DDR4-2133
Asus GTX 980 Ti Strix OC
MSI GTX 980 Ti Gaming 6G
Samsung 850 EVO 500GB
Corsair Carbide 600Q
Fractal Design Define S
SilverStone Strider Platinum 750W
Scythe GlideStream 120
Phanteks PH-F140HP x 2
Retail prices are subject to fluctuations. Please
use the shopping links to check on current pricing; don’t rely on the
The total for our build comes out to US$1845 though you can cut that
down by as much as US$500 by selecting components that offer more value
without sacrificing performance. This is still a healthy chunk of change but
it’s probably close to the minimum you want to spend for a decent 4K gaming
experience. Even with this seemingly high-end configuration, image quality settings
have to be scaled back significantly on the most demanding titles to get playable
framerates. That being said, gaming at 4K with low detail levels can still be
breathtaking at times just due to the sheer number of pixels, something that
maxing out the eye candy at 1080p simply can’t replicate. It’s especially well-suited
for slower paced games where you can take time to admire the scenery. Building
for 1080p and allowing the display to upscale is also a viable option, though
how well that works likely depends on the display. It looked pretty darn good
on the Vizio M43-C1 when used in this fashion. This combined with its low lag
gaming mode makes it an excellent gaming TV.
If this level of performance is not satisfactory, you’ll have to wait for the
next generation of graphics cards for something better to come along, at least
if you’re determined to stick with a single GPU configuration. Adding an additional
GPU (or more) means the budget motherboard and power supply options are out
of the question, and the increased power draw has tremendous ramifications on
the noise level. It would be impossible to maintain the same noise output on
load as our build without employing some serious cooling. Liquid cooling would
likely be necessary which means having to deal with a noisy pump or two, spoiling
the superb idle acoustics. Our system produced just 14 dBA@1m when not being
taxed, with fans running at minimum speed. The 23 dBA@1m attained during the
gaming tests is not as low as some of our previous guides, but this configuration
is substantially more powerful and it’s still very quiet, at least by gamer
standards. Also, the type of sound it emits is inconspicuous, such that background
music being played at a moderate volume is enough to cover it up completely,
which is as good as one can hope for with any gaming PC.
* * *
Articles of Related Interest
Corsair Carbide 600Q Inverse Tower
Asus GTX 980 Ti Strix OC Graphics Card
Asus Skylake Z170 Motherboards: Maximus VIII Gene vs. Z170-A
Quiet SLI Gaming PC Build Guide
Quiet Liquid Cooled Gaming PC Build Guide
Quiet Mini-ITX Gaming Build Guide #3: BitFenix Prodigy Edition
Basics & Recommendations
* * *