ppietra

robaba said:

frantisek said:

inexco said:

To me the headline implied running Windows on the Air.
I was expecting to see Parallels or another emulator running Windows better.
My 2012 Mini runs Windows better than most dedicated Windows machines.

You all should know that Roseta will smoke any other x86 emulations for long as M1 has unique features speeding x86 emulation. dont remember what exactly but was something with memory ordering or task scheduling....

Rosetta is NOT emulation but code translation from Mac(Intel) to Mac(ASi) that generally takes place at installation time.  AFAIK MS chose to do a more traditional run-time emulation which generally sucks in comparison.

even if MS does port Windows to ASi it will still lack native software unless the individual developers cross-compile.  Would they do that for such a small target audience if they aren’t being forced to do so is the question.  Here’s where Microsoft gets screwed by its primary selling point—backwards compatibility.  They will have to sacrifice that if they want to make a forced-march to a different ASI, and they must make a forced March if they want their developers to go along.  My money says they never leave x86, for better or worse.

it will be interesting to see what solutions Arm is able to bring to the table, but they are even more compromised by their architecture.  

From what I understand MS also uses code translation with recompilation, but it does that in small batches of code when needed, while Rosetta usually recompiles all executable code together ahead of time. MS emulation seems to cache translated code but I imagine that method still creates additional lag compared with Rosetta. I also imagine that Rosetta is far more optimised considering that Apple made the M1 and Rosetta together and knows pretty well what it wants, while MS probably has to worry about emulation method compatibility with future ARM chip variants.

I think MS problem isn‘t being unable to force-march developers, it’s not giving good incentives for developers to program for ARM. There is no strategy for a broad ARM ecosystem of Windows computers, from tablets to workstations. If there aren’t ARM machines for power users and there is no Windows ARM machine that really disrupts what Intel machines offer, then most developers will not see why is it worth investing time and money supporting these ARM machines, even more so when there is x64 emulation.
I mean, if no one tries to sell an ARM Windows machine arguing it has better performance than an Intel alternative, then no consumer that buys it will really care too much about wether the app is native or not... it’s not like they will have high expectations.

chadbag said:

We’re the Mac benchmark tests running under Rosetta 2 ?   If you run the Windows on Arm under x86 emulation you should at least compare oranges to oranges and run x86 Mac versions under Rosetta 

True! PCWorld just "recycled" the MacBook results from MacWorld, didn’t bother to make a good analysis.
Anyway, the Geekbench results are all native.
In Cinebench the MacBook app is running native, but we can use the Mac mini Cinebench under Rosetta scores from Anandtech: 999 and 5257 (single and multi). Still miles away.
Edit:
 Martin Nobel run Cinebench R23(x64) in Windows inside a virtual machine running on a Mac M1. The single core score was 495!!!!!!! Still ahead even though it is running in a non-optimised virtual machine. Amazing.

This technology is also based in research developed in Portugal, one of the lead researchers involved is actually from a Portuguese University. I wonder if those licenses involve the portuguese contribution.

mknelson said:

How would Apple differentiate between the driver and a passenger using Facetime safely?

It can’t without some serious artificial intelligence, nor can any other cameraphone. THat is why Apple has only implemented this with Carplay - contrary to what is assumed this has actually been implemented