I recently took two PC case fans, cable-tied them to a Radeon RX 580, and tracked how the card performed.
As soon as the publish button was hit, there was feedback from our tech-savvy readers: you should turn the fans around.
In the initial testing, the fans were positioned to suck hot air off the card’s heatsink – as seen in the image below.
This was done mainly because turning the fans around would expose their power cables and plastic backplates to the heatsink.
As the heatsink gets very hot (it’s why it has “heat” in the name) there was a concern about the plastic or power cables melting – or at least being damaged.
The second attempt
Following the feedback, I cut the cable ties, turned the fans around, fastened them in place, and tested the card again.
Despite feeling like a noob who had put the fans on the wrong way the first time around, it is always better to take advice and try it out with the hope of improved results rather than sulk.
The new setup can be seen in the photos below.
Careful consideration for the fans’ power cables was taken, and the layout of the heatsink meant that the cables could be positioned in a way which did not result in them touching the metal.
This would not only allow for better cooling, but also removed the risk of the power cables getting too hot.
The next step was to put the new card in my gaming PC and run the benchmarks again.
The photos below show the card – and the fan hub which powers the two fans via the motherboard – installed in the PC.
For the test, I ran the same benchmarks as before.
I first used a stock RX 580, and then the revised “custom cooling” RX 580 to run several Hitman 2 benchmarks.
For the benchmark, the following settings were in place:
- 2,560 x 1,440 resolution.
- Level of detail, texture quality, shadow quality, and reflection quality – all on high.
- Texture filtering on Anisotropic 4x.
- Motion blur and dynamic sharpening on medium.
- The test level was Miami.
The benchmark was run several times on each card to ensure the results were consistent, and at the end of each test Hitman 2 provided an average frames per second result as a performance measurement.
HWMonitor was also used to monitor the GPUs’ temperature, with the peak temperature per card noted during each benchmark.
The results, detailed in the table below, clearly show that turning the fans around and blowing air towards the heatsink – instead of sucking it off – improved the performance of the card and lowered its maximum GPU temperature under load.
|Cards||Hitman 2 Benchmark||Max GPU Temperature|
|Stock RX 580||45.40 fps||71 Celsius|
|Stock RX 580||45.86 fps||71 Celsius|
|Stock RX 580||45.95 fps||71 Celsius|
|Custom RX 580||42.09 fps||86 Celsius|
|Custom RX 580||40.03 fps||87 Celsius|
|Custom RX 580||43.02 fps||87 Celsius|
|Custom RX 580||45.17 fps||79 Celsius|
|Custom RX 580||45.99 fps||81 Celsius|
|Custom RX 580||45.38 fps||81 Celsius|