M2 Macbook

Has anyone got RPCEmu to work on an M2 MacBook (with Rosetta)?

On an M1 Mac Mini, RPCEmu-Interpreter runs (albeit a bit slowly) while RPCEmu-Recompiler is so slow as to be almost unusable. I would expect an M2 machine to be much the same, but by the sounds of things it’s a whole lot worse :-(

Have you tried both the Interpreter and Recompiler versions of RPCEmu?

Presumably it could just be recompiled to run on the M2?

Presumably it could just be recompiled to run on the M2?

RPCEmu-Interpreter: Perhaps.

RPCEmu-Recompiler: Definitely not. The code assumed that it was running on Intel silicon.

It runs well on my M2 MacBook Air with only very occasional crashes. Networking works automatically. The only real problem is that the mouse jumps around in an unusable way in full screen mode so that’s not possible despite the option being there. I’m running RISCOS 5.28. The “monitor” is set to an Acorn AKF60 which limits the resolution but things broke when I tried to change it to anything else – I may be missing something. I have the “Reduce CPU usage” option on as I’ve noticed rpcemu having a big impact on battery life on laptops in the past – the M2 is so fast that even with that and the interpreted version RISCOS is as smooth as anything I’ve used it on. The version I’m running was downloaded from:

https://github.com/Septercius/rpcemu-dev/releases

Has anyone got RPCEmu to work on an M2 MacBook (with Rosetta)?

I did my own release, the interpreter runs fine on my M1 Max. I also made an experimental version that is compiled for M1 and uses Rosetta for the recompiler. The code works, memory bandwidth is good, but video performance are bad, so probably it needs more work.

There is an ongoing effort to build a recompiler for ARM, but not sure at which stage is it at the moment.

If you want to have an idea of the performance of my own port, I listed them here (expect better performance on M2):

https://www.riscosopen.org/forum/forums/10/topics/17008#posts-129504

I did not measure Septercius port.

HTH.

I had a go at writing an AARCH64 recompiler a year or so ago, I didn’t get very far.

rpcemu-recompiler

That’s a build if anyone’s interested in it. It’s just slightly under double the interpreted speed. It’s also probably buggy. (It attempts to recompile one instruction I think…. badly…. :D )

Someone else on here says they get a buzzing on boot, however I didn’t find that with the stock image I was using.

Also, just to be 100% clear, it’s very hacky, mid being played around with code. I wouldn’t normally release code in this poor state, it’s only there because a licence file told me to do so. ;)

I need to find a use for that..

Why did you push the buggy code to git?
Um, a license file told me to??

Um, a license file told me to??

Probably something like the GPL that says words to the effect of “if you release this program to the public, you must also supply the source to build it”.

He had, so he has. ;)

Probably something like the GPL that says words to the effect of “if you release this program to the public, you must also supply the source to build it”.

I mean it’s pretty much that. I had a binary I thought others might find interesting/useful, although I’ve not tested it other than booting RISC OS and running a benchmark.

The code is licenced under the GPL, so to save me having to keep a copy of the source code lying around for the next three years it’s much easier just to include it in the DMG and point out it’s poor quality. :)

@Mr Rooster :
https://github.com/merryhime/dynarmic

would this help to improve the AARCH64 backend?