I agree with this, but for reasons other than failure rate. The arctic cooler just isn’t good at mechanically securing SP5 and providing enough pressure to the socket evenly.
All the other SP5 coolers have a sturdy plate that covers the entire IHS area of SP5 to give even pressure while Arctic’s solution relies on unsupported heatpipes to give pressure to the “long” edges of the CPU IHS:
to arctic’s credit, they provide an .STL file of an adapter to mount standard 120mm fans to the Freezer 4U M which I’m 99% sure is the same as the Freezer 4U-SP5
If you ever have the free time (the benchmark would probably take ~3 hours), I would be super curious how these CPUs perform the CFD-EM benchmark:
The only way I’ve been able to get any of the modern Intel Xeons to perform well on Windows is to manipulate the core bitmap in BIOS so Windows uses the “right” cores. It’s not as intuitive as you’d think because the two different CPUs in a dual socket system can have different available bitmaps so you might need to give each CPU a unique bitmap.