Say hello to OSX on X86

[HOM]

There seems to be a new PowerPC emulator available called PearPC.

Some user has gotten Panther (http://www.osnews.com/story.php?news_id=7085) to run on it. He claims it's slow as hell, but the screen shots look legitimate. Now, I know that Apple's EULA prohibits running OSX on a non Apple computer, but can Apple really enforce that if this picks up speed?

More importantly, if Panther on PC becomes all the rage will this force Apple to sell an official version of OSX for X86?

If this is fast enough on a top end PC would you switch to PC?

[alcimedes]

i believe what you're referring to is PearPC, not pearl.

also, the processor side of things runs approx 500x slower than the real deal, so it's not too much to worry about now.

with Apple's excellent integration of hardware and software, i don't think they have to worry too much about this, at least not at this time. it might run on a PC, but you really wouldn't get the whole experience, not even close. although, the clones sold pretty well when they were in production, so what do i know. :D

[_Ω_]

I think this line sums up the only positive aspect of this feat:

It took, in total, more than 5 hours. And that was just the first cd. Since I had nothing else to do (PearPC took 99% of my processor and all the RAM it could possibly find), I actually started to clean my bed/computer room. Thank you, PearPC.

[stoo]

I'm installing Panther on PearPC at the moment. 500x slower is for the "regular" version. The JITC version is much faster.


It's not just about speed. Currently PearPC doesn't support network cards or physical optical drives under Windows.

[Wickers]

But that will change... with all this press the project will get quite the boost.
Now could someone build a chip that does the conversion from PPC to x86... add it to your PC using that nifty and fast PCIexpress interface... that should speed things up a bit.

[Luca]

\/\/ickes... something like this (http://www.totalimpact.com/products/total_mpower/total_mpower.html) maybe?

http://www.totalimpact.com/graphics/mpower.jpg

[thuh Freak]

Now could someone build a chip that does the conversion from PPC to x86...
well, not a hardware sol'n, but someone did come up with an ingenious idea to bridge the 'cross-platform' problem. they are called high level languages. ;) if osx was completely oss, you could, in theory, cull out the proc specific stuff, rewrite it for x86, and compile. infact, i suspect that most all the ppc specifics are available in darwin's src.

[Brad]

Moving to Mac OS X! :)

The biggest problem with emulating PowerPC on the x86 architechture and why it will never be as fast as VirtualPC (x86 on PPC) is that aside from a different instruction set, the PowerPC also has many more registers. PPC has four times as many general purpose registers than x86 (32 vs 8). So, to get any decent speed, you're going to have to get some assembly programmers out there on the project with genius skills that have never been seen before.

[stoo]

But x86-64 has sixteen 64bit general purpose registers, which are logically the same size as 32x32... :D

Has anyone got this working in Virtual PC yet? I'd love to see a screenshot.

[123]

Moving to Mac OS X! :)

The biggest problem with emulating PowerPC on the x86 architechture and why it will never be as fast as VirtualPC (x86 on PPC) is that aside from a different instruction set, the PowerPC also has many more registers. PPC has four times as many general purpose registers than x86 (32 vs 8). So, to get any decent speed, you're going to have to get some assembly programmers out there on the project with genius skills that have never been seen before.

On the other hand, besides registers, x86 operands can be fetched directly from memory (less registers necessary) and internally, those processors use many more registers anyway. Also, I doubt that VPC's x86->PPC register mapping is optimal, I suspect it doesn't really use much more PPC registers than the 6+2 x86 registers + additional registers to load/store operands/results (at least for the actual code) but I might be wrong and they apply some smart register renaming techniques. Still, the eventual speed difference between VPC and PPC->x86 emulation is probably not as big as could be assumed from the register files. (that's at least my opinion after seeing how far they (Intel et. al.) got with their horrible architecture, a miracle all by itself)