Published on 24th March, 2017
AMD’s new 65w R7 1700 Ryzen CPU would have been the ideal candidate for this build, but we are still waiting for a Mini-ITX AM4 motherboard to be released onto the market. So far only Biostar, Gigabyte and ASRock have made any kind of announcements for Mini-ITX AM4 motherboards, but these are at least a few months away from release.
With lack of AMD Ryzen ITX support, I instead opted for Intel’s current fastest 65w gaming CPU, the i7-7700. This was combined with an MSI B250i Gaming Pro, its one of the cheapest 200 series boards currently available, but it actually has a really nice set of features.
For graphics, a GTX 1070 is currently the most powerful GPU available that is true ITX length. Zotac have recently released a GTX 1080 Mini, but it measures 210mm in length so its not quite ITX and will not fit inside the LZ7 (max GPU length 185mm).
For CPU cooling I would have liked to have gone with the Scythe Big Shuriken 2 again, with its 10 heat pipes and 120mm cooling fan, but it was out of stock at my local shop! They did however have a Noctua NH-L12 which is renowned for being an excellent and quiet LP cooler.
Removing the top fan of the Noctua NH-L12 results in a cooler height of 66mm, 3mm shy of the LZ7’s absolute maximum CPU clearance (under 60mm cooler height recommended for coolers with top mounted fan). Online reviews show that even with just the bottom fan this is quite a capable cooler and exceptionally quiet, so I thought I’d give it shot, especially considering that low noise full performance is my primary objective.
Unfortunately, I was not able to install the cooler with the fins running perpendicular to the 140mm side case fan which I was hoping for, the 92mm fan mounted under the heatsink fouls on the RAM modules. This only leaves 1 possible orientation to install the cooler without it overhanging the motherboard edge:
The heatsink clears the Corsair Vengeance LPX RAM with a bit of room to spare, the fan is set to pull air through the heatsink downwards (default orientation):
Motherboard installed into the case, looking good:
GTX 1070 installed, you can start to see all the clearances, it’s going to be cozy:
I felt that the NH-L12 heat pipes were sticking out a little too close for comfort to the backside of the GPU, the gap was about 2mm. I slipped this ram module packaging tray over the end of the heatsink to provide a shield to protect from electrical shorting, it fit perfectly:
Power supply installed, not much wasted space in this build! The free space in front of the PSU is essential for cable routing with this modular Corsair SF450 power supply:
A small rubber pad was inserted under the PSU to help keep that 3 – 4mm gap:
Rear of the case showing CPU cooler height clearance, the Noctua NH-L12 fits with only a few mm to spare:
The Gigabyte GTX 1070 ITX OC is the current king of <180mm length graphics cards, it’s fairly tall though. Here it is shown next to a Gigabyte GTX 960 4Gb OC (top left) and an EVGA GTX 1060 SC (top right):
A side view f the case showing the Cyclone Vent pattern, these cutouts are designed to reduce fan turbulence and noise, providing more airflow and lowering system temps compared to traditional slot or perforated hole style intake vents:
The Gigabyte GTX 1070 ITX is now assembled into the case and ready to go. The case in the background is a half assembled Silverstone RVZ02B:
Rear view of the case showing the arrangement of the various components, this photo also highlights the 360 degree GPU ventilation which is a key part of the case design which gives the case such good GPU cooling potential:
Next up, how did the finished build perform?
MSI’s BIOS handles fan control very well, I was able to set both the CPU and System fans to run at less than 500 RPM when idle, which was practically inaudible even with my ear right next to the PC. This is the first motherboard I’ve used where the fans keep turning when set below 30% power, in fact I was able to get the Prolimatech Vortex fan down to a stable 350rpm at around 20% power!
Running the fans at such low RPMs does mean that idle temps are a little higher than they could be, but personally I prioritise a silent system over absolute lowest temps.
For the purpose of this testing I set the idle speed of the fans low enough for them to be silent but with some room to ramp up and still remain quiet, this resulted in fans speeds of around 550rpm at idle. The GPU and PSU fans were at zero rpm.
To stress the system a combination of software and games were used to test different real world scenarios, the system was also tested fully stressed with Prime95 which pushes the CPU beyond what it would normally see.
GPU COOLING SETUP
The Gigabyte GTX 1070 ITX OC graphics card has a thermal target temperature of 82C, the card automatically applies the maximum clock speed it can achieve (effectively overclocking itself) until this temperature is reached. Once 82C is reached the GPU will begin to ‘back off’ the clock speed and increase fan speeds to maintain a stable temperature. For my particular card this boost speed goes as high as 1840Mhz in open air, thats a 300Mhz increase above its 1531Mhz stock clock, and even 110Mhz above its advertised boost speed of 1721Mhz.
For the first round of gaming tests the ‘silent’ profile in the Gigabyte GPU monitoring software was used, this gives the GTX 1070’s 92mm fan a soft power curve which tops out at around 1550 RPM when it hits the 82C thermal target.
Play close attention to the GPU clock speed results, this shows you how much your graphics card is ‘backing off’ in order to maintain its thermal target.
For the second round of stress testing, the GPU will be tested with both ‘Silent’ and ‘Auto’ fan profiles to see what kind of difference it makes.
CPU COOLING SETUP
Again silence is a priority, using the MSI monitoring software a gentle fan curve has been used, opting for low RPMs until the CPU reaches 60C, the curve then climbs steeply up to 100% fan power at a CPU temp of 82C.
This fan curve is also mirrored with the Prolimatech Ultra Sleek Vortex 140mm case fan, prioritising low RPM’s with a CPU temp of below 60C and ramping up the speed as the CPU starts to reach 70C.
This combination of fan curves results in a silent system under normal usage, but it also does a good job of keeping the system running quietly under moderate usage such as 1080p gaming.
RESULTS – GAMING
Games were played on a 1080p monitor with graphics set to their absolute maximum settings for each game including full Anti-Aliasing. Each test was carried out for a minimum of 30 minutes allowing temperatures to stabilise, the results shown are maximum temperatures recorded during each test.
The tests were carried out in a very quiet room with an ambient noise level of 21dB, at idle the noise meter could not detect a noticeable increase in noise level at a distance of 30cm from the PC, to my ears the system was silent with it sat on the desk next to me.
Starting up Unigine Valley, this test puts a high load onto the GPU with an uncapped frame-rate taking the GTX 1070 up to 98% average usage. The i7-7700 however was not stressed much at all with a CPU usage of only 21%.
With the high GPU usage it reached a temperature of 82C with a fan speed of 1531 rpm, after 30 minutes of testing the GPU clock speed was running at 1747 Mhz, not quite its full boost but still above its standard boost clock of 1721 Mhz which is not bad going considering the fan profile is set to ‘silent’.
The system noise rose by 3dB from idle, this increase in sound is coming from the GPU which is the only audible component in the system whilst running this test.
This game is allot of fun to play, I’ve been a fan of Battlefield ever since the original 1942 game, I think this latest version brings back allot of the games original charm mixed with the faster paced action of the more recent releases.
Battlefield 1’s graphics are incredibly optimised, this is evident with only 70% of the GPU’s potential being used even with the graphics maxed and running at a stable 60 frames per second. On the CPU side the game pushes the i7-7700 up to 48% average usage during a multiplayer session, this is the highest of the games tested today.
The moderate GPU usage means the GPU only reaches 77C, this is 5C below its thermal target and therefore the card maintains its full boost speed of 1835Mhz with a fan speed of 1490rpm throughout the test.
In terms of noise the GPU is the only component in the system that can be heard in the silent test room, the noise is like a faint whooshing sound, it can be heard when sitting next to the PC, but from more than a 2 meters away it is virtually inaudible.
With 48% average CPU usage during the test the i7-7700 does reach 60C, but the fan curve is set to run under 700 rpm up to this temp, at this speed the CPU cannot be heard under this level of usage.
I think this shows that the Noctua NH-L12 in combination with the Prolimatech Side fan is more than capable of providing virtually silent cooling for a 65w TDP Intel processor under ‘normal’ gaming conditions inside an LZ7 case.
This game looked STUNNING, the GTX 1070 did a great job of providing incredibly smooth and high detailed graphics, it was a joy to test.
The Witcher 3 pushed the GTX 1070 up to 89% GPU usage but only 27% CPU usage, very similar results to the Unigine Valley test. However, this game caused the GPU to emit some coil whine more evident than the other games tested, this whine was the loudest part of the system and could be heard above the fans. As you can see from the results the sound level rose to 26dB, this increase was from the coil whine and NOT the fans.
Don’t get me wrong though, 26dB is still incredibly quiet for a PC system under gaming load, this test was carried out in a silent room and noise levels measured at a distance of 30cm. The coil whine would not be loud enough to be noticed over normal game volume.
RESULTS – STRESS TESTING
Software tested for minimum of 30 minutes per test, results shown are maximum temperatures recorded during each test.
As you can see Prime95 pushes the CPU up to 82C maxing out the CPU fan to keep a steady 81-82C. At this point the CPU fan can becomes audible, but again, it’s not loud, just a low level humming from the Noctua fan. Bare in mind that Prime95 is above and beyond the load your CPU is likely to ever see in normal conditions, and remember, the gaming tests did not push the CPU beyond 60C which is a more realistic real world test.
Prime95 + Unigine Valley
Starting Unigine Valley at the same time as Prime95 saw some minor increases in temperatures, the system temp rose by around 4C caused by the extra heat produced by the GPU running at full load, this in turn pushed the CPU temp up by 1C which isn’t too bad considering. During this test the Noctua CPU fan running at 100% (1442rpm) was more audible than the GPU fan (set to silent fan profile) running at 1531 rpm. This is probably due to Noctua fan experience turbulence pulling directly through the heatsink fins, the GPu fan is a little less restrictive.
Changing the GPU fan profile to ‘Auto’ increases the GPU fan power quite a bit, resulting in a fan speed of 2055 rpm and a fully stressed temp of 69C, this is a drop of 13C compared to when running the ‘Silent’ fan profile. At this point the 2055 rpm GPU fan starts to become louder than the Noctua CPU fan running at 100%. The lower temp also meant that full GPU boost was maintained throughout the test, my particular card automatically boosts up to 1835 Mhz, a full 300 Mhz above stock clock speed!
Personally I don’t think the increased noise from the GPU fan is worth the extra 100 Mhz of boost clock compared to using the ‘Silent’ fan profile which runs the same test at a stable 1747 Mhz, this is still an impressive boost more than 200 Mhz above the advertised 1531 Mhz stock clock.