I currently play it on Steam in bootcamp but was concerned about running it when I eventually get a new Mac.
I heard that the Gog store version will run under Crossover (or Parallels) but I was not familiar with vcmi.
Such a thing probably won’t come from Crossover or Parallels, but by one or two innovative programmers (just like how most breakthrough innovations happen), but someone needs to code a comprehensive Windows/x86 emulator or “virtualization instance” for Apple Silicon Macs, where ingenious programmers do a
LOT of GPGPU optimization coding (as much as humanly possible!) to get it to
approach speeds of a native Windows/x86 machine (like the Mac used to be).
I’m often reminded of Steve Wozniak who used to obsessively revisit his Assembly language or “machine language” code to see if there were
any possibilities for him to tighten up the code or eliminate steps to save a processor cycle or two or to spare a memory register or two, to make his app run the same only more efficiently, less resource-intensive and
faster.
(And always bear in mind, when you’re thinking about the amount of RAM in
any recent Apple device — be it an iPhone or a Mac — it’s
shared RAM; shared between the CPU and GPU.)
I don’t know enough about GPGPU coding to know if a programmer
knows for certain that
no more serial tasks can be performed parallel on GPUs instead of the CPU in a given piece of software — or — if there is
always room for more GPGPU improvements to be found upon close enough inspection.
This doesn’t even include other blocks on Apple silicon like the Neural Engine, AMX coprocessors, ISP, ProRes engine, etc.
(And, yes, I do realize that they all use different data orders/structures/operational functions/context switching, but for serial CPU instructions that
are suitable— if they
can be performed
faster — I wish there were programmers genius enough to maximize their use for general purpose instructions — to make a fast Windows/x86 compatibility environment for Apple Silicon Macs.)
If it’s the case that, for instance, the ProRes engine doesn’t
have to be used specifically for what it’s “supposed” to do, but can be tasked with instructions for non-image-related, non-codec, non-graphical-purpose vector and floating-point arithmetic
faster than the CPU cores, it should be exploited in cases where speed is
such an important consideration — like a Windows/x86 compatibility environment.
This all requires a lot of
very unconventional thinking for any normal coder — but foreign platform compatibility environments are a wildly unconventional undertaking on their own.
Maybe this will have to wait for M3s.
I’ve read that the A17 Pro SoC is within reach of Intel i9-13900K and Ryzen 9 7950X processors in single-core performance — and those are
DESKTOP processors!
I can’t
imagine what the performance will be on the forthcoming desktop class
Apple M3 SoC!
(I just hope not to be disappointed.)