Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Apple's Rosetta uses some special hardware in the M1 and later chips to make some x86/x64 translated instructions much faster. I believe it was the load and store instructions.

Microsoft doesn't have access to these instructions for their own implementation (because they are not ARM standard), but theoretically they could have qualcomm or other SoC providers add some custom hardware that does the same. But in a fragmented landscape that can be hard, some windows laptop SoCs might have it, some might not.



Microsoft could still work with their partners to implement this. Yes, the PC market tend to be fragmented, but what I assume would happen is that some ARM PCs will have the hardware acceleration, and for those it would lead to a much better performance. But, for the ones without hardware acceleration, it would just fallback to software emulation.

And as time goes on, more and more OEMs would probably favor adding the hardware support because people would prefer to buy the ones with faster x86 emulation.


Rosetta on the M1 enables the total store ordering from x86, which has a stronger memory model than that of ARM. It also can compute the parity flag in hardware <https://bytecellar.com/2022/11/16/a-secret-apple-silicon-ext...>.


If I remember right, total store ordering is actually a standard ARM extension, and not something Apple came up with on its own. I'm not sure about the flag computations, I suppose those are custom?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: