On this page you can find the temperatures of both the SSD-controller as well as the nand-chips after two custom runs in Atto Disk Benchmark. For this test we used transfer sizes 1-64 MB, a total length of 32 GB and a queue-depth of 8 in order to put the SSD under heavy load. First we take a look at the temperatures with next to no airflow, as is usually the case in practice. The second chart shows the results with the airflow of a 92mm-fan.
Of course the temperatures are at their peak without heatsink: we measure a controller temperature of 111 degrees. This is unhealthy for your hardware and leads to a serious performance-impact, as we will show on the next page. Out of all of the heatsinks the MSI M.2 Shield clearly offers the worst performance. This is not entirely unexpected, because it is only a thin piece of metal. MSI does proof that they can create a good heatsink, because the more luxurious M.2 Shield Frozr achieves one of the best results in this test.
The best cooling comes from the combined chipset- and m.2-heatsink included with the ASUS X299 Strix-motherboard. Note that this heatsink is also the biggest one by far and in practice there will be more heat sources connected to it, like the chipset. For this test we only mounted it to the SSD, like all of the other heatsinks.
With the airflow of a fan the temperatures go down quite a bit. Across the board we see a decrease in temperature of about 20 degrees, but the differences between the multiple heatsinks remain. There are barely any changes to their order of performance: the MSI M.2 Shield remains the least effective, while the EK heatsink, the MSI Shield Frozr and the ASUS TUF and Strix perform relatively well. The difference between the most and least effective heatsink is exactly 10 degrees, while this was 23 degrees without fan.