Thermal Throttling

Due to the compact form factor, M.2 drives lack the ability to cool themselves and usually have to rely on passive airflow instead. All vendors include some form of thermal throttling on their drives as a safeguard, which limits throughput once a certain temperature is exceeded.

On this page, we will investigate whether the tested drive has such a mechanism, how high temperatures get, and what effect this has on performance. We will test the drive in a typical case and the M.2 slot between the CPU and VGA card. A second data point shows the result with a 120 mm fan directly blowing on the tested drive. Each of the charts has time moving from left to right, with the blue line displaying transfer speed in MB/s and the red line showing the temperature in degrees Celsius (measured using SMART).

Results from this test setup are not comparable to our 2019 SSD bench because we're using a different case and a CPU cooler that generates some airflow around the CPU socket.

Reads

Writes

There's quite a bit of throttling with the uncooled drive. Even the SSD with a fan blowing directly onto it sees a little bit of thermal throttling.

I've been wondering whether the "graphene" heatsink is worth anything or whether it might hinder cooling by trapping heat, so I removed the heatsink and ran just the bare drive without any fan through our write test.



As you can see, there's now much more throttling than with the heatsink installed, which confirms that the heatsink does have a positive effect on cooling. However, drives like the ADATA Gammix S70 are using the same controller with better heatsinks that mitigate throttling much better.

Thermal Image & Hot Spot

The images from the thermal camera confirm that the heatsink spreads out the heat, so it gets dissipated over a larger area. The thermal reporting of the drive is also quite accurate, at 87°C measured controller temperature vs. 85°C reported in software. On other SSDs, we've seen differences of up to 30°C here.