Hi everyone - like many of you, I love the x86 vs arm race that is going on and actually use an M1 Max as my daily driver - with a 5950x + 3080 with 32GB of b-die as my backup/gaming machine - and so I'm always on the lookout for faster and more capable arm boards, or even full servers.

Sadly, most arm servers are still much too expensive to use in my homelab, although the Honeycomb LXS Server is getting closer to reasonable prices ($3k for the server, $800 for just the board + cpu).

A friend of mine left Apple’s processor division and went to work for Qualcomm and is holding out hope that the Qualcomm Cloud AI 100 will provide some competition in the server space and also believes that their desktop efforts will bear fruit. Their desktop effort are based on a company they acquired Nuvia and not ‘yet another’ re-purposed mobile chip. They claim we’ll see silicon shipping in 2024, but we’ll see.

I wrote all this today because I was interested to see how my 6-core ODroid-N2+ SBC (widely regarded as the faster ARM SBCs out) fared in Geekbench against my 10-core M1 Max - I know, not fair at all, but still interesting.

(And yes, I know Geekbench is seethingly hated amongst professional reviewers, but it works for our purposes just fine …)

So, without further adoo:

Device Single & Multi Core
O-Droid-N2+: 357, 1190
M1 Max: 1791, 12721
Ryzen 9 5950X: 1706, 17604