Testing & Results
Unlike the way I test CPU coolers (by trying to isolate a single, tested item), “case testing” is a loosely defined term in my reviews. Generally, I’ll just make observations along the way, since the results are generally specific to the exact configuration I used for testing (making it next to impossible to reproduce the results…therefore not very objective or relevant). I’ll commonly build a different system into each different case, as I try whenever possible to build something that matches the case (not much point in placing an ITX board in a full tower, etc.). Be wary of comparing my impressions with objective data, since my results are decidedly unscientific. Hopefully, my impressions combined with your planned build will help give you an idea of what you could potentially expect.
I’m always interested to see how well smaller chassis can handle hot enthusiast-class hardware. I generally won’t push anything too hard, but I at least want to know if a particular chassis will actually cool a typical gaming build. I’ll start with the configuration pictured above, and let the case warm up with a few gaming or benchmark runs. After that, I’ll run through Unigine Heaven or 3DMark to simulate a consistant gaming environment, and see if I can get an idea of what someone could expect with similar hardware.
Tested System Configuration:
– Core i5-3570K @ 4.2 GHz
– 2x4GB DDR3 1866MHz CL10
– AMD Radeon R9 290 (945 MHz Core, 1250 MHz Mem)
– Samsung EVO 250GB SSD
– Rosewill Lightning 80+ Gold 800W PSU
– Integrated fan controller on high setting
Since the Cooler Master Nepton 280L is already installed and I’m still working on a 24/7 CPU overclock for this machine, I’ll be primarily interested in the GPU results. High-end GPUs like the R9 290 consume at least twice the power as Intel’s Ivy Bridge processors, so the graphics cards will be the major culprit of heat anyway. The ambient temperature varied between 24C and 24.4C during the tests, and internal case temperature measured at the front panel (with the Samsung EVO’s temperature sensor) was 33C.
In the Node 804’s stock configuration, a 3DMark 11 benchmark run resulted in temperatures of 71C for the GPU and 60C for the CPU – this was with the Nepton 280L’s two 140mm fans on the lowest possible setting, so better CPU performance would easily be obtained (the JetFlo fans get LOUD though!). The nice part about two chambers is the ability to almost completely isolate the CPU temperature from the GPU temperature with the use of a radiator. As I mentioned before, I’m primarily concerned with the graphics card temperatures.
In the “Stock +” configuration (with the right rear 120mm moved to the front intake of the main chamber) I saw slightly improved temperatures of 70C for the GPU (and 59C for the Core i5). Incidentally, the additional intake fan must have cooled the SSD a bit since the sensor on the EVO dropped to 32C during this 3DMark run.
Adding a spare 140mm Fractal Design Silent Series fan as a top exhaust in the main chamber didn’t change the story much – the graphics card still hit a temp of 70C. I actually noticed this with the top exhaust fans on an Arc Midi R2 (it uses a primarily mesh top cover as well) – most fans don’t have the static pressure to shove air through the relatively dense filter. The Silent Series R2 fans do a better job than some, but you may want to find some fans designed to be mounted against a restriction for this area.
The Heaven benchmark is primarily graphics-intensive, it doesn’t use excessive CPU/physics like the 3DMark 11 benchmarks. Consequently, the R9 290 ran a bit warmer during that run (73C), while the CPU was much cooler (52C).
Finally, while a mining enclosure this is not, I had to see how the Node 804 would perform under a full GPU compute load. 87C (with a quick spike to 89C) was about the maximum temperature the aftermarket-cooled R9 290 reached – even adding a fan pulling air from the bottom of the chassis didn’t change temperatures more than a degree. I’d be interested to see how some Noctua NF-F12s or Corsair SP120s would perform against the top mesh, and maybe that would help exhaust the heated air faster.
While I originally ran out of time for some Crossfire results, I was still curious if the direct cooling method that was so effective for the Carbide Air 540 would work in this smaller chassis. I finally got a chance to throw two XFX Radeon 270X graphics cards in the Node 804 to see what type of temperatures they’d produce. Again, I’m mostly concerned with GPU temps; the CPU’s radiator is in the other chamber and positioned as an intake (and is essentially independant of the heat on this side). The below temperatures were all achieved with the case fans on High and the fan configuration shown in the photo below (the additional 140mm Silent Series R2 exhaust fan remains from the last test).
To heat things up right away, a full GPU compute load across both of the 270Xs (running at 1050 MHz, memory at 1400 MHz) resulted with top GPU temps of 82C and bottom card temps of 73C. That’s about the same result as some full tower enclosures! The recently reviewed Thermaltake Core V71 full tower kept the same cards (on a different chipset and board, remember!) at 85C and 61C, although the ambient temperature where I tested that case was MUCH cooler (14.3C vs 23C). That’s a pretty impressive result for such a small chassis regardless, and the setup was actually a bit quieter than the fans on the R9 290 running at 100% during the same test.
A few 3DMark 11 benchmark runs told the same story – 63C top / 59C bottom, with a few rounds of Battlefield 4 confirming it (63C top / 57C bottom). It seems my suspicions were not far off; even open-air coolers like the XFX Radeon R9 270Xs used here were manageable in the Node 804. That’s pretty uncommon for micro-ATX enclosures.