Corsair Carbide SPEC-04 Review: Fan Placement Testing, Noise, & More

By Published May 19, 2017 at 1:14 pm

Additional Info

  • Component: Case
  • Original MSRP: 50
  • Manufacturer: Corsair



The SPEC-04 is the first case to receive our recently-improved, finalized benchmarking treatment. We still performed the torture test that we usually publish (generating as much heat as possible), and although that’s great for comparing relative performance between cases, it’s not a real-world scenario. We have therefore added our standard Blender rendering test (performed on the CPU and the GPU) to simulate a realistic production workload, and the 3DMark Firestrike Extreme stress test to simulate load-intensive gaming workloads. We also took the liberty of performing some tests with an additional fan installed in the case.

This makes the SPEC-04 one of our most in-depth case reviews, perhaps adjacent to the NZXT Manta review.

Torture Test

spec 04 torture cpu

In our torture test, CPU dT got almost exactly as high as it did with Be Quiet!’s Pure Base 600 (with the top closed). Temperatures reached about 100 Celsius at times and, by the end of the test, the CPU was throttling in an attempt to keep temperatures down. The stock SPEC-04 isn’t the hottest case on the bench by this metric, but with the CPU beginning to throttle, it couldn’t have gotten much hotter.

spec 04 torture gpu

GPU dT was even worse in comparison to its peers: 63.9C dT over ambient is a higher temperature than any case except the Pure Base 600 with its fans turned down, and that includes the completely fanless In Win 303 (which breathes through the bottom, permitting the GPU to bring air in where the SPEC-04 bars this).

spec 04 torture cpu frequency

Yikes -- those frequency versus time charts are a little scary, but remember, this is a torture scenario. The other cases can handle it, but our $50 DUT with one fan doesn’t hold up. We need to leverage our new real-world test cases to better understand performance; this is exactly the reason we began introducing those tests. Before getting to those tests, let’s add a fan to the case.

Additional Fan Torture Testing

Corsair’s 120mm stock fan may be decent, but it’s pushed to its limits trying to move air through a case on its own. For our final tests, we did what the SPEC-04 really deserves: we added an additional ~$8 Arctic F12 fan. Filling every fan slot would give the SPEC-04 a huge boost, but it would also add ~$20-$30 to the price, at which point Corsair has better cases that we’d recommend -- like the 270R. For our first experiment, we placed it in the top-front slot, freed-up by the lack of optical drive support, and set it to maximum RPM.

spec 04 fan placement cpu

Starting with CPU torture test temperatures for the extra fan tests, we see that adding a fan to the front grants an immediate performance uplift of about 10C, reducing temperatures to 57.8C delta T over ambient. This helps dramatically in CPU temperatures, and brings us down under throttling territory; unfortunately, the fan is partially obstructed by the front panel, so its ability to improve thermals is limited by this. The rear exhaust configuration helps marginally more – just outside of our test resolution – and can be explained by ridding of radiative heat off the GPU backplate, in addition to being fully unobstructed. Either way, adding one fan to the case helps tremendously, and would be a worthwhile investment.

spec 04 cpu frequency fans

Here’s a look at CPU frequency versus time with these configurations. With the stock configuration, we can see clear throttling as the CPU enters the 90-100C range of our torture workload. Adding a fan in either configuration removes this throttle event.

spec 04 fan placement gpu

spec 04 extra fans gpu freq

Looking at GPU temperatures after adding an extra fan, we see a reduction from 64C delta T over ambient – which is roughly the throttle point – to around 60.5C delta T with an extra rear fan, or around 60C with the extra front fan. Performance between these two is insignificantly different, meaning that they are within our test-to-test variance and resolution accuracy. The two are effectively the same for GPU performance, and we cannot claim a definitive advantage between front and rear mount options. That said, we can show a definitive improvement with an extra fan over the stock configuration. A fan mounted in the lower position would further aid the GPU, though the CPU would be left to run hot.

Blender Renders

Still, this is a $50 case, and $50 cases shouldn’t necessarily be expected to compete in a thermal torture test with more expensive cases that ship with more fans. That’s why we added these new tests, after all.

Blender rendering was the first of our two more realistic tests, and temperatures were far more reasonable. Our methodology going forward will be to do two renders, one using the CPU and one using the GPU, but in this case we did three. When rendering on the CPU, we tested leaving the GPU fan speed at the standard 55% or turning it down to 25% to minimize any effect on airflow, and we will be using the 25% method from now on.

[blendercpu_gpu25% cpu temp vs frequency]

[optional: add blendercpu_gpu55% cpu temp vs frequency to show consistency between tests]

With CPU rendering enabled, we see the Corsair Spec-04 with a single fan pushing the CPU to 43.5C delta T over ambient with our cooler. This is about 6.1C warmer than the open Thermaltake Core P3 configuration, which runs a load CPU dT of 37.3C. Idle temperatures are also a few degrees higher on the GPU, which isn’t doing work in this scenario, but is humming along in the box.

spec 04 blender cpu worker

spec 04 blender gpu worker

Moving to Blender GPU rendering, we’ve only got two cases on this chart as it’s still new – but it’s a good first look. The Corsair Spec-04 with its single fan at 1300RPM runs a CPU dT of 35C over ambient, with the GPU load dT at 37.6C over ambient. The SilverStone RL06 at 1000RPM, with its 3x 120mm front fans and 1x 120mm rear exhaust fan, is able to sustain a 22C dT CPU temperature and 24.7C dT over ambient GPU temperature. For an extra $15, that’s a significant reduction in temperatures.

spec 04 blender gpu freq temp

Rendering on the GPU again lacks a truly fair comparison, as Silverstone’s case came with three more fans installed. Frequency did remain fairly stable, steadily dropping only about 40MHz over the course of the test and staying close to 2000MHz overall. This is within normal behavior of Pascal cards.

3DMark Test

spec 04 3dmark stress

3DMark is one of our new introductions to case testing. We’re using a looping FireStrike Extreme test for half an hour here, producing load on both the CPU and GPU in a less torturous manner than is done in our extreme burn-in. This more closely resembles gaming workloads for extended periods of time. As these are new tests, we are still adding to the charts, and only presently have two devices plotted.

The Corsair Spec-04 runs a CPU temperature of 38.9C delta T over ambient in the 3DMark burn-in, which is about 9C higher than our open-air Core P3 configuration, effectively the resting state of where you’d be mostly without an enclosure. Or a limited one, anyway. The GPU temperatures run about 5C higher, with a delta T of 59.4C on the Corsair configuration. To get a better idea of how this looks in reality, let’s look at a frequency versus temperature chart like we use in our GPU reviews.

spec 04 3dmark freq v temp

In this chart, we can see how GPU frequency responds to rising temperatures. Our fan is fixed to a 55% fan RPM, which is around where most GPU fans configure their maximum profile before sacrificing frequency rather than noise. Technically, you could run a higher GPU fan speed and solve for this frequency throttling problem, but you’d be drastically increasing noise – something to the effect of a 2x perceived noise increase, depending on what video card is used.

Regardless, there’s a clear down-trend in core frequency until a point where it fluctuates between the high 1700s and low 1800s, resultant of an 84C maximum desirable GPU core temperature. The VBIOS begins throttling back performance to sustain this temperature target under stock configurations, which means losing framerate due to heat. And again, this isn’t a torture workload – it’s just 3DMark.

Running a lower-end GPU like the 1050 would let you get away with the warmer case, but a stock or $20-$30 CPU cooler would still call for an extra fan when the system is under 100% workloads.


It’s not just about lower temperatures, it’s about lower noise. Sure, 60-70C might be acceptable and won’t throttle too hard outside of torturous scenarios, but if you can run lower fan RPMs on all internal components, that means lower noise.

The SPEC-04’s single fan isn’t incredibly loud, but the limited size of the front panel intakes combined with a double layer of mesh gives it a disproportionately loud whoosh. It’s not a problem at normal fan speeds with only the stock fan installed, but our experiments with two front intake fans at 100% duty cycle got noticeably noisy.

corsair spec04 noise

The Corsair Spec-04, by nature of running a single fan, is able to get away with a lower noise output than some of its neighbors. The case doesn’t make any attempts to contain noise – especially with that front mesh intake – but it also doesn’t emit much, given system components that are reasonably inaudible. We measured about a 36.7dBA output with the Spec-04, placing it effectively equal with the 270R at 1300RPM and SilverStone RL06 at 1000RPM. The SilverStone RL06 is able to cool significantly better even at this reduced 1000RPM, and does so while retaining a similar noise output. That said, it’s also an extra $15, so it just comes down to a cost argument and to how much noise and thermals matter to the user.

Conclusion: Corsair SPEC-04 vs. 270R, RL06, & More

corsair spec 04 case 3

The test results might seem negative, but any issues that the SPEC-04 has are perfectly normal at its price. The basic features are adequate, but with the added benefit of a side window and an unusual appearance (if that’s your thing). There is some air filtering and there are plenty of available fan mounts, and we’d strongly recommend filling one of them, as the additional front intake slots are what give the 04 real potential to perform perfectly adequately for a $50 box; however, the main marketing point is the unusual front panel: for customers who love it, this is an easy choice. For those who don’t, well, it’s still an easy choice. Aesthetically, this is a more affordable version of the SPEC-ALPHA, and anyone who’s been tempted by that but unwilling to pay $70-80 should look into the 04 (although the ALPHA is currently on sale). For those who don’t, there are slightly higher-end cases packaged with more fans, like the windowed Corsair 270R (two fans, better performance in our tests) or the Silverstone RL06 (chart-topping performance, but louder). We’d recommend either, although they’re more expensive, they also don’t need any additional fans installed to perform well.

At $50 plus a ~$7 fan, you’re looking at $57 for the Corsair SPEC-04 -- not bad, and it becomes an acceptable, balanced performer once that extra fan has been dropped in. Just get the fan. It’s worth it. For $57 with the extra fan, we feel confident recommending the case for budget gaming PC builds. If two extra fans enter into your mind, unless the face of the SPEC-04 is needed, just jump ship and go with either the 270R windowed ($70) or SilverStone RL06 ($75).

Writer, Testing: Patrick Lathan
Video Host, Test Lead: Steve Burke
Video Producer: Andrew Coleman

« Prev Next

Last modified on May 19, 2017 at 1:14 pm

  VigLink badge