dpJudas wrote:Keep in mind that while the M1 doesn't use a fan in the mac mini and mac book air, it does use one in the mac book pro. How fast can it go? We'll see. I'm certainly keeping an eye out for anything high end from Apple that's for sure. But at the same time, while it does win big time in the same class it is competing in, in some of the tests I saw it also lost against certain mid-range AMD CPUs. That's why I'm saying it is difficult to fully conclude where we go from here. I think we still need more data.
All I say about that is: This is the first chip of this family. It cannot be expected to be the ultimate killer.
No, it surely does not magically beat all x86 chips out there - but what it does is being equal with my 8 year old desktop system on GZDoom, which still got a competitive CPU, with a passively cooled SoC, essentially meaning an entry level chipset with integrated GPU. If it manages that, we can surely expect a lot more once the platform matures and new avenues of improving performance are found, especially if others start competing with Apple on this turf. This was merely the first shot in the upcoming CPU war and it was a direct hit right into the center of x86's stronghold.
I guess we'll see how Intel will play it - but long term I have no illusions that x86 may come out the victor in this game - it's far too old and crusty for that.
Rachael wrote:That is interesting.
Well - apparently the M1 Macs still have Boot Camp. So if you get sick of Mac OS ... there's still always an alternative.

Though - to be quite frank, I expect that drivers will be far better developed on Linux at this point than Windows for ARM, but that is sure to change in the future if Microsoft's ARM Windows offerings can at any point compete with Apple's.
I wouldn't count on Microsoft here - but since Windows for ARM is not locked down to these, there's a very good chance of other manufacturers betting on ARM as well. Only then we'll really see how this thing will play out. Change will come, that's for sure, though. We now have proof that ARM is good enough for desktop work.
Rachael wrote:There's a misconception that there's no hardware acceleration other than Direct3D for Windows ARM at this point - which from what I have seen simply isn't true. What's really missing is actual driver support. But at the OS level everything is nearly identical to x64, other than having an emulation subsystem to support x86/x64 compiled programs.
As far as Vulkan goes - unlike with OpenGL, Windows does not actually provide the backend for that. The GPU driver does, directly. I'm sure to remain standards compliant the GPU's Vulkan dll has to call up ICD's like OpenGL does - but Microsoft provides no Vulkan support at all, whatsoever - neutered or otherwise.
Obviously the last work hasn't been spoken here - but I'd expect two developments here:
1. Turning OpenGL into a pure wrapper API.
2. Overall less direct reliance on native APIs.
I'll openly say that once I find the time to do it, I'll consider migrating GZDoom to some low level wrapper engine to decouple myself from direct interaction with the APIs and streamline the backend. I do not think that native Vulkan support is the way to go - the entire API is far too fragile and hostile for me - better offload the work to people who feel more comfortable with this kind of work. Getting D3D and Metal support through one of these would far outweigh the drawback of adding in an external dependency.