Is it possible for someone to make a reverse Rosetta instead of PowerPC to Intel, Intel to PowerPC?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Reverse Rosetta for PowerPC
- Thread starter Anthony-Cic
- Start date
- Sort by reaction score
Running Snow Leopard on a PowerPC Mac
Some interesting background reading here: https://web.archive.org/web/20041109090012/http://www.wired.com/news/technology/0,1282,64914,00.html 80% speed of target architecture - that would've been nice :) Here's the guy that did it:
forums.macrumors.com
I think the intent was to run later Intel Mac software on PPC.Running Snow Leopard on a PowerPC Mac
Some interesting background reading here: https://web.archive.org/web/20041109090012/http://www.wired.com/news/technology/0,1282,64914,00.html 80% speed of target architecture - that would've been nice :) Here's the guy that did it:
the problem is twofold - emulating the instruction set and providing the expected library support. The former would be like QEMU (or the FEX-Emu project that supports Linux x86/x86_64 emulation on 64-bit ARM). The latter is the bigger problem if you are trying to support Mac applications that expect newer OS X or macOS releases - the PPC libraries don’t have the newer functionality.
Al
It's probably much less of a hassle to just get a cheap Intel Mac.
Last edited:
Virtual PC
No Mac software, though, and it's pretty darn slow.
No Mac software, though, and it's pretty darn slow.
@bunnspecial - @Lastic got Tiger for the DTK to run in QEMU on a G5 at a snail's pace, so if you somehow got, say, Snow Leopard to run and had infinite patience things could get interesting.
Last edited:
It might be possible but you have to remember that Intel processors at the time of the switch were much more powerful than the PPC ones they supplanted. That meant you could run PPC executables on top of Intel and the power of the Intel processor could suck up the overhead of code conversion and still leave an acceptable user experience. The same won't be true of running later Intel binaries on earlier PPC hardware. Look how poorly VPC ran once you tried to run a contemporary OS and applications, even on a G5.
Years ago, there was a tool for POWER7 or POWER8 computers to run amd64 Linux software, and just recently the ppc32le Void build came out which also allows you to run i686 Linux software, but that's just Linux software and you have to have a Linux host. The closest I can think of would be installing Darling (assuming it builds for i686) and/or GNUstep on top and doing that, I heard they were working on getting ppc32le working on Power Macs but I don't know if that's still on what with the ppc32/64 big endian discontinuation.
Virtual PC
No Mac software, though, and it's pretty darn slow.
Yes, if only curious that's the choiceIt's probably much less of a hassle to just get a cheap Intel Mac.
It's probably much less of a hassle to just get a cheap Intel Mac.
Or for free due to people abandoning early Intel Macs, especially with the arrival of the M1 generation.
While not OS X-compatible, I did set up QEMU's Linux userspace emulation and managed to run x86 executables in a chroot.
I might revisit this.
I might revisit this.
Well, almost. The Mac Mini 2009 seems to be averaging around £35 on eBay now. I just put Monterey on mine yesterday, and it seems to run fine enough for a C2D. It will be more than adequate for SL.Or for free due to people abandoning early Intel Macs, especially with the arrival of the M1 generation.
I remember someone releasing a working proof of concept of a qemu integrated in cocoa that allowed seamless emulation of Intel apps on PPC. At the time it only worked on very simple apps and I don’t think it went much further, but theoretically it’s totally possible.
Tell that to Adobe CS2 on my 1.4 GHz C2D MBA: unusably slow.
To be fair, I have that MBA and everything on it seems horribly slow. ULV chips probably weren't the best for much back then.Tell that to Adobe CS2 on my 1.4 GHz C2D MBA: unusably slow.
You could probably compile/run QEMU.
Good luck with performance.
Seriously though your best bet for PPC software with any level of performance these days is probably linux and compiling stuff native for PPC.
Good luck with performance.
Seriously though your best bet for PPC software with any level of performance these days is probably linux and compiling stuff native for PPC.
Last edited:
The point I made with my link was that the technology was there in 2004 to translate X86 code at 80% efficiency but it was never deployed/endorsed for Apple computers - why would it be? Apple wouldn't be especially keen for users to extend the life of their PowerPCs.
No market for it.
Still no market for it really, only people running 15+ year old hardware because they refuse to buy new machines (and thus are also very unlikely to pay what is required to develop/market a viable product).
This software development malarkey isn't free in terms of time/effort/money. Unless there's a return on investment it ain't happening unless a hobbyist does it, and even then... small target audience, likely almost all of which do not include developers capable of writing it.
Things are slightly different now as the retro scene (whether it be computers or anything else) is a thriving market - as evidenced by the amount of expensive peripherals enthusiasts buy to interface their ancient computers with the modern world.
A developer creating some must have software at a price to make it worthwhile isn't so much of a stretch.
Yup.
Lets say it takes 1 man year of development to get a viable product off the ground (likely 2-5x that as this is by no means a trivial project, its not just glue code calling libraries... but... for the sake of argument). For someone qualified in the requirements for this that's say, $100-150k or more in dev cost (wages he could have got doing something else) that needs to be recouped. Plus - he needs to spend that time up front. No money incoming until the project is sellable.
How big do you think the market is, and how much of the market will this guy be able to capture? Are there really a few thousand PPC Mac users willing to pay $100 each to even break even? I'm not so sure. What about in 3-5 year's time (in case it takes that long to actually complete)?
And that's before you consider ongoing product support. Multiply the numbers by X years depending on how quick a dev could produce it... but I really think 1 man year is extremely optimistic - based on my experience with in-house software developers at work for projects far less complex.
Stuff like Rosetta is NOT cheap to make and not cheap to support. There really aren't many people skilled enough to develop it - intricate knowledge of both PPC and x86 at a hardware programming level required. This is why even Apple cut off support eventually, even though they spent the time and effort to develop it in the first place.
TLDR:
It's a multi-hundred thousand dollar high risk project with dubious return prospects. Anyone with the funds would be better off just sticking the money in the bank, indexed fund, crypto portfolio, etc. Or churning out a heap of simple little iOS games and maybe lucking out with the next angry birds.
Could a hobbyist do it out of love? Maybe.... but again, the skills required here really aren't common.
Last edited:
Replace your worst case scenario with a best case - equally plausible.
My take wasn't particularly referring to a translation layer but "must have software" - in my opinion that would be an efficient web browser.
£35 Buy it now? Fair play to you if you picked one up for that price because it was a bargain. I've just checked eBay UK and I can only see a couple of MM 2009's at £30 ish and that's for the opening bid. Everything else ranges around £60 to £90.
I was thinking of much earlier machines than that though: such as the 1st and 4th generation A1181's that were gifted to me by someone who knew I'd find some use for them as opposed to seeing the pair continue to go neglected in a closet.
With the more recent Intel products, Apple saw to it that their lifespans would be restricted at the point of production. Offering computers that could continue to be expanded and used for decades: that just won't do.
Not to mention that running something written for, say, Intel Leopard is something, but newer and newer apps will require more and more features available only on more recent versions of macOS, so you’d have either to implement those or to have the whole Intel OS running as well. Sounds to me a really daunting task.