Fortunately, I’ve recently got access to M1 Max (not Ultra, though), so if you have any test program to run, I will try to run it and measure the timing. Hopefully, the test programs will be as “minimal and simple” as possible, such that the bottleneck is clear (e.g. CPU load vs data transfer), but other types of codes may also be interesting.
By the way, there are a lot of sites that compare “CPU performance” of different processors, e.g.
- PassMark CPU Benchmarks - New Desktop CPUs Performance
- PassMark CPU Benchmarks - New Laptop CPUs Performance
Here is also a comparison between M1 and Xeon E-2186M (the latter used in another thread)
I am wondering to what extent I can use these comparisons for specific type of calculations… (e.g., CPU or memory intensive). Looking at some other sites, M1 and M1 max are just similar to, e.g., Ryzen 9 series for some calculations, which is of course good but not “super” fast (as suggested by the bandwidth). I’ve also run my codes, but the experience was similar. So, the comparison may depend very much on the type of calculations.