The Zalman CNPS2X with AM4 adapter
Zalman released the CNPS2X back in July 2013. Now, they have an AM4 bracket for it – thus expanding its capabilities! That also means this is the perfect time to revisit my tiny machine.
Back in March of 2017, I built a rig with the intention of having an HTPC capable of gaming that could double as a grab-and-go stream/game PC for events. The idea was ‘it has to be small and powerful’, which means small and hot generally. The original build used a Noctua HN-L9A, a 65w TDP cooler.
My goal this time was to increase the cooling potential while keeping the low volume case, which means limited CPU cooler choices. When I saw that the Zalman CNPS2X (a 120w TDP cooler) had an AM4 adapter kit, I thought it would be a shoo-in for the job.
In the Box
The first thing you’ll notice out of the box is that it is ‘no frills’. While the box itself has a blue and white color scheme, the descriptive text is a bit heavy on the words. But you’re not buying this for the box.
Inside, a clear plastic tray holds a couple mounting brackets, a few screws, a pouch of thermal compound (enough for about 3 applications) and of course the small copper cooler.
The cooler is a throwback to the late 90’s CPU coolers: solid copper (color), circular, and a smokey translucent fan. The cooler being circular is nice in the fact that you don’t have to worry as much about exhaust direction. On square-ish coolers, the fins are stacked in such a way that air travels north-south or east-west (as you look down on it), whereas this exhausts air 360 degrees around. So you don’t have to worry about VRMs or other motherboard components creating any sort of real airflow obstructions. The base of the cooler seems a little small, but we’ll get to that later.
Test System and Methodology
- CPU: Ryzen 5 1600x
- GPU: Radeon RX 560
- RAM: Geil Evo X (16GB)
- SSD: Crucial MX300 (1TB)
- CASE: Fractal Design Node 202
I decided to test it out and use AIDA64 as my bench test of choice. To create a nice heat box, I also made sure to have a load on the GPU and Memory.
The baseline for my comparison, the Noctua, held 74°C after 60 minutes. I ran this test 6 times, allowing the system to return to the room’s ambient temperature of 68F (21C). For each test, the Noctua held at 72°C. Now having a good baseline test, I dismantled and cleaned the CPU and cooler to remove the thermal compound.
Time to test out the Zalman. I applied Arctic Silver (the same compound used on the Noctua) to it. To my surprise, the Zalman – having nearly double the theoretical cooling capacity – performed worse, reaching 82°C.
Thinking there had to be something wrong with the seating, I dismantled and inspected it. The spread looked fine. So I cleaned and applied the new thermal compound to try again. Same results, time after time. I switched to the included thermal compound in the CNPS2X box and found worse results yet at 82°C.
It’s all theoretical
My theory is that since the Zalman cooler’s base only covers some 60 – 70% of the Ryzen 5 1600X’s IHS, it provides less cooling. Compared to the Noctua (which had full coverage), the CNPS2X could not pull the heat away fast enough. I really wanted to reduce heat, though, and at this point, I had run out of the thermal compound.
So I decided to pick up some liquid metal by way of Thermal Grizzly Conductonaut (WARNING: DO NOT USE ON ALUMINIUM). This helped immensely. Running the same tests, the Zalman CNPS2X now ran at 72°C, which was 2°C lower than the L9A, and 10°C cooler than the previous tests of the CNPS2X both using Arctic Silver. In the spirit of fairness I decided to try the Conductonaut on the Noctua L9A, and saw a similar 11°C drop in thermals, coming in at a mere 63°C over the course of my testing.
A cooler’s primary function is, well, to cool. But I also want to mention noise. Noctua is renowned for their fans’ acoustic performance, and the Zalman wasn’t far off the mark. While I don’t have any scientific equipment to get precise decibel readings, the Zalman only felt a hair louder under load and had a little bit of a whine. At idle, however, they were nearly indistinguishable.
Let’s talk money: the Noctua [$39.99] is about 10$ more the Zalman. Both require an additional AM4 kit, though it seems you can buy an updated Noctua with the AM4 bracket included for the price, whereas you’ll have to hunt for the AM4 bracket for the Zalman.