This article was written by Steve Burke & Patrick Lathan from GamersNexus
This question gets asked every year, it seems, and is worth a revisit: Where in the case should I put my liquid cooler? Top, front, or rear?
Ultimately, the answer is a big “it depends” – but we still tested it. Case testing is one of the least exact sciences, because it’s simply impossible to control for every possible variable. The case alone has a huge impact on thermal performance, next dictated by the fans and coolers within the system, then the video card (how do its fans behave?), the CPU cooler, all the way down to the power supply. Every single component impacts internal case thermals, and that means that the best anyone can do is create a system analog that reasonably represents most builds of a certain type. For radiator testing, we simplify matters by eliminating the various air coolers designs – liquid coolers ultimately use parts that are easier to compare. In the very least, we can look at liquid cooler placement performance in a few different cases. This includes measuring how much a front radiator might reduce CPU temperatures at risk of raising GPU temperatures, or how a top-mount might do the opposite.
All fans will be configured to 100% in our testing.
|Cooler||Kraken X42 (140mm radiator)|
|CPU||Intel i7-6700K @ 4.4GHz|
|GPU||MSI GTX 1080 Gaming X (OC Mode)|
|Motherboard||MSI Z170A Gaming M7|
|Memory||Corsair Vengeance LED 32GB 3200MHz|
Three cases with unique qualities were chosen for this test: the S340 Elite (relatively normal layout), the SilverStone Redline 06 (unusually high airflow), and the Corsair 600C (inverted motherboard). This was not a comparative test between the cases; comparisons and opinions on the cases can be seen in the individual reviews we’ve published at GamersNexus.
A note on charts: two configurations are compared in each chart, and the lower temperature between the two is highlighted green for each component. This does not mean that the difference is significant, it just means that one number is lower.
This article will just focus on the NZXT S340 Elite. We sort of had to hack the solution together, seeing as our VRM heatsink and RAM wouldn’t allow a top-mounted radiator (X42) to clear. Some cases would allow this more readily.
Here’s a look at the configuration, though we obviously used a radiator (X42) instead of the air cooler:
Results for the S340 were consistent with the RL06, which has the same conventional front intake, rear/top exhaust system. CPU temperatures were predictably lower when the radiator was placed in the path of unwarmed in-flowing air, while GPU temperatures were lower when inflowing air was not warmed by the CPU. The two configurations tested are more directly comparable than they were in the RL06, since the direction of airflow did not significantly change. In the S340, results were consistent between load and idle testing.
Silent Pump Radiator Orientation Testing
With the pump set to “silent mode,” which establishes a pump speed based on the liquid temperature, CPU deltas should have been at least slightly higher with other temperatures less affected. This held true except for the top rear configuration under load, where dT was not only a degree lower than the front upper configuration, but also lower than the same test with the pump at full speed. There were no anomalies in ambient temperature, CPU load, frequency, or voltage, but differences this small could be attributed to test variance, especially with the added complication of a pump speed curve.
We have other cases tested in our article on GamersNexus, but the foundation is this: It depends on the case and the cooler. In the instance of our S340 Elite configuration, we couldn’t properly fit the radiator top-mounted (our RAM and VRM heatsinks didn’t allow clearance); realistically, that forces a front-mount, and so thermal differences really become irrelevant. For some of the other cases we tested, we generally found that front-mounted radiators always raise GPU temperatures as the CPU heats up. This makes sense, since the front-mounted radiator is dumping heat onto the GPU. CPU temperatures rise a little bit if the cooler is rear-mounted, but there are a few important things to note here: One, CPUs are way more tolerant of temperature fluctuations than GPUs, and also have significantly higher thermal ceilings; two, GPUs won’t dump heat straight into the CPU cooler (generally speaking), so the impact of the CPU temperature on the GPU is more profound than the inverse.
Ultimately, we generally recommend trying to keep the GPU out of heat paths as most have a temperature ceiling of 82-85C prior to clock throttling – and GPU clock throttles are more noticeable in gaming performance. That said, as long as there’s adequate cooling within the case (and the video card has a decent cooler), throttling shouldn’t really be a huge concern. Just mount it where it fits best. If multiple options are present, and if “best” cooling is desired, consider front-mounted for the best CPU cooling or top/rear for the best overall balanced cooling.