Successful PearPC/Mac OS X Install Documented

rocketjam writes "OS News has an article by a user who successfully installed Mac OS X using the 0.1 version of PearPC, the PPC emulator for x86 machines. He said it took 5 hours to run the first install CD but he did get it up and running on an AMD Athlon XP 1600+ with 512MB of RAM. The article has several screenshots of the Mac OS X install and new user set up running on his machine." See our previous story.

Comments

I can see myself using this

[magefile]

I like Mac OS X (Darwin + nice GUI + It Just Works(tm) for most stuff), but I hate the way the mice and keyboards feel. Plus, you can save on hardware by using x86 stuff ...

How efficiently does it run? I.e., how fast/expensive a box do I need to get a normal experience?

Re:I can see myself using this

[TravisWatkins]

Nope, even on top of the line PC hardware its slow as dirt. But there are already people working on ways to speed it up. Expect massive speed increases in the near future.

Re:I can see myself using this

[Anonymous Coward]

and each layer having two virtual child machines... or three, or more... (somebody please build an über-powerful cray and do this :P)

Re:I can see myself using this

[Kazymyr]

Here's an idea: why not put a PowerPC chip on a PCI card and use that to run the instructions natively, with the emulator front-end being a wrapper for the hardware (and possibly provide the rest of the emulated system)?

Like older macs used to have a PC compatibility card.

I always wanted OSX on PC

[stealth.c]

But does this thing perform similarly to a similarly powered PPC?

Sigh. If only Apple would port the thing themselves. Add a windows compatibility layer and you've got one hell of a competitor to Microsoft.

Re:I always wanted OSX on PC

[Lord Crosis]

...and only Steve can decide if it's worth the trade off. Personally I think Apple stands to profit more from software sales than they stand to lose from hardware sales. Apple's hardware (especially their laptops) is innovative, and that will continue to a large portion of their sales.

Even if MacOSXIntel is in the works I think Apple needs to do some things before they can consider going toe-to-toe with MS. If MS sees Apple as a threat they would logically pull support of all their products from Mac OS X. This means that Apple at the very least needs a viable alternative to MS's biggest non-OS products: IE and Office. Apple already tackled IE, and Safari is great. They don't have anything that can compete with Office, though Keynote could be taken as an indication that this is the direction they are heading.

There is a slightly dated, but never-the-less relevant opinion column by a friend of mine, Joshua Thorpe, on my website at http://www.macopz.com/columns/jt/thinkswitch.html

-=(Lord Crosis)=-
Andy Rooney of Borg: "Ya ever wonder WHY resistance is futile?"

hmm

[Anonymous Coward]

Isn't one of the biggest pluses of a Macintosh system the flawless integration with the hardware? That's always been something I've admired, and something that's been a pain in the butt for both Linux and Windows. I wonder how stable this runs?

Re:hmm

[drinkypoo]

Flawless my ass. Ever owned a blue and white G3 and decided to upgrade it? If it's revision 1, then you cannot use UDMA transfer modes on most hard drives, and have to resort to PIO. You can use slower UDMA methods on some drives, and some will do the whole shebang, up to whatever UDMA mode it supports most. Unfortunately mine was not a drive which you could use and if there was any significant CPU use whatsoever it would write invalid data. I verified that this was the cause of my woes with an OS9 app that tests disk writing and yes indeed, I had this problem.

There is a workaround which was considered acceptable given that these are some slow-ass macs, which is to use the PIO modes. However, you need a third party disk driver to do this. The cheapest software I could find to work around the problem was $80.

And of course, there's no firewire booting on those models, so I couldn't get around the problem that way, either.

Apple has since suppressed information about this by removing the applicable documents from the techinfo library when it was folded into their current support system. I have only excerpts from the document.

Now, I can forgive apple for having a bug and for not replacing motherboards. Well, almost on the second count, but certainly I will forgive an error, even though Sun managed to use the same chip in several Ultra systems quite successfully. But what's stupid is that the OS was not designed to address this issue in the hardware.

Apple's support of their own hardware is selective and short-lived at best, as evinced by the lack of support for several macs with G3 processors in OS X. The fact that you can make it run on them with third party software that tricks the installer into going ahead and doing its job is particularly pathetic.

The biggest plus of a macintosh is that it is friendly and generally consistent in behavior. Macs are workhorse machines which will not always be the fastest horse but will usually run for a long time. My mother used her Mac IIci with System 7.1 or something for absolutely ages, until just a couple of years ago in fact. She paid five grand for it when it was new (and worth eight, or at least, it sold for eight grand with a two page mono and an 8*24 display card) and she definitely got her money out of it. I bumped up the hard drive (to 2*200MB!) and the ram (to 40MB) while she had it, never even did a cache card (by the time they were cheap, she was more or less done with it) and she used pagemaker, illustrator, and photoshop throughout the system's life, and her work has won several awards in the process. Current macintoshes are basically the same; somewhat quirky, mostly reliable, and quite consistent. And, still very pricy. But, if you get more work done on a mac, it's worth more money, and some people certainly don't seem to get as much done on windows as they do on a macintosh.

Re:hmm

[merdark]

Integration does not mean upgrading hardware beyond what Apple themselves will do. I just got my first Mac, and it's far far more integrated than any PC laptop I've ever used. I don't expect to be able to upgrade it much though, but that is a separate topic.

What you are complaining about is the Mac's life cycle and lack of upgrades. Both are valid concerns, but neither has anything to do with Macs having good software/hardware integration.

Re:hmm

[WinterSolstice]

I know what you mean!!!

I had the worst time putting AIX 5.1 on these old RS/6000s we had laying around. Sure, they were about 4 years old, but that's ok, right? It's still a RS/6000!

Sheesh. When you get stiff vertical integration, you get *stiff vertical integration*. We have systems here that literally must run the same OS they shipped with. And they were millions of dollars. I understand that you want to have the new OS on the old hardware (which is typical in the PC world) but that's why there are minimum requirements. In the case of Apple, they rebated a lot of software for this sort of problem. They didn't really have to. It was just to try and make customers happier. Heck, IBM would have simply laughed at you if you bought ZOS for a machine that wouldn't run it. Then offered you a new lease :)

PS - I'm not apologizing for Apple, I just think that people whine too much about this. Ever tried to upgrade a Commodore? How about an OS/390? Macs are purpose built machines, not like x86 boxes. If you buy one, deal with it.

-WS

Awesome...

[stuffman64]

the emulated processor is about 40 times slower than the host processor.

Great, if you were to do this with a 2GHz Pentium, you would get the performance equivalent of around 50MHz. There is no way in hell that OSX would run decently at that speed, what with all the transparancy and animation of the UI. But hey, at least it works.

what I don't get is...

[bennomatic]

if Mac OS X runs on emulated generic PPC hardware, what's to stop people from running MacOS X on any number of PPC platforms? What does Apple do to "squash" that sort of activity?

Re:what I don't get is...

[tepples]

If Mac OS X refuses to install on a computer that doesn't report as having been manufactured by Apple Computer, then telling an implementation of OpenFirmware to lie about the maker of the system would probably not infringe Apple Computer's copyrights or trademarks. Sega v. Accolade, 977 F2d 1510 (9th Cir. 1992).

Mostly a Proof of Concept

[gbulmash]

As noted in the article, this is 0.1 version software. As well, it runs on top of Windows. That makes the "40x slower than the normal processor" claim seem somewhere about right.

Now, the tricks as I see it are:

• Optimize the code in future releases for better speed.

• Build a version that boots out of Linux or BSD to minimize overhead. Seems that since OSX is PPC BSD at its heart, there might be some sort of way to lower redundancy if the emulator were running in x386 BSD.

Re:Legality

[exp(pi*sqrt(163))]

But it's not a real computer using the software. I mean, nobody things that my spaceship in defender is a real spaceship do they?

Re:Slashdot condones piracy?

[HeghmoH]

There is no indication that EULAs (an unsigned "contract" that is dictated by only one party and can't be examined before purchase) are legally binding, and certainly breaking an EULA is no major sin. If he had a purchased copy, it's certainly not "piracy" even if it is illegal to break an EULA.

Re:Slashdot condones piracy?

[SiMac]

No, it doesn't say Apple hardware. It says Apple-LABELED hardware. You could stick your PC board in a Mac Classic. Hell, you could even just slap an Apple sticker on the side and it would be legal.

Aren't loopholes fun?

Why just run OSX?

[Orion Blastar]

Why not Yellow Dog Linux for PPC, why not AmigaOS 4.X, why not MacOS 9.X, why not the PPC version of BeOS? Anyone tried those yet?

OS X machine to ssh in and compile on?

[molo]

This could be helpful for developers looking to test their open source code on Mac OS X.

Does anyone have any OS X machines available for open source developers to use? Something ssh-able with apple's developer tools (make and gcc) would be sufficient.

If no one knows of any services like this, would any OS X people be willing to open up user accounts on their boxen? (PearPC or real hardware, either would be fine) email me: molotov1134@hotmail.com

Thanks,
-molo

Re:Slashdot condones piracy?

[Kadagan AU]

Actually, I'd worry more about the legality of his Windows 2003 license. There is no proof behind this, but I'd imagine that if it's a home system, it wasn't actually purchased. If you're going to pay that kind of money, plus buy a copy of OS X, why not just buy a Mac? From the article (with emphasis added):

Test system:

- AMD Athlon XP 1600+;
- 512 MB SDRAM;
- Ati Radeon 9000 with 128 MB DDR-RAM;
- CMI-8738 based 5.1 soundcard;
- MSI K7T Turbo2 mainboard;
- 40 GB harddisk;
- Standard ps/2 keyboard;
- Microsoft Trackball Optical (USB);
- Windows Server 2003, Enterprise Edition (host);
- PearPC 0.1, emulating Mac OS X 10.3 Panther (client).

It hits a specific economic bracket dead-on.

[solios]

Because hypothetically, this thing will get optimized to the point where it should be possible to run OS X acceptably. And there are people out there who are interested in such a thing, such as myself- I recently broke the bank to acquire a dual G4 450 for 500$- and it took another 300$ in upgrades to make it useable (to say nothing of the ~200$ worth of parts I'm permaborrowing to make it functional for entertainment purposes). That's a four year old machine.

By contrast, I can get a used PC (from a coworker) that's faster (133mhz bus as opposed to the 100 in the G4), at a used price of half the present value of the parts he put into it... which is about 160$.

The economically disadvantaged don't get the luxury of modern high-powered Macintoshes- for the price of a three-year-old G4, I can build a CURRENT PC.

If I could run OS X at useable speeds through an emulation system on a CURRENT PC, I'd buy the hardware and do things that way- seeing as how a current PC (bare bones) is between 1/4 and 3/4 the price of a current useable (re: expandable) Mac.

Re:OS X Panther Here

[bluekanoodle]

While I have yet to hit the 10 minute mark, if you don't choose all the extra fonts, and printer drivers, the OS X install is surprising zippy. On both my 12 inch powerbook and 15 inch i can do a nuke and pave in about 15 to 20 minutes. That sure as hell beats my XP box's 45 minute install times. My Suse Box is sitting here reinstalling now and its pushing an hour, but then that's with almost everything.

Bearing in mind Pear PC is only at v 0.01

[Phil John]

and is currently running only 40 times slower than host, that's very impressive given the register starvation problem. With future versions I'm sure they will be working on optimisations, the graphics code may be slowing things down simewhat as I understand Quartz uses 3d graphics hardware for some of its compositing magic.

I think this is definately a project to keep an eye on, plus with platforms like Athlon64/Opteron this may be far more viable.

Picture this: Pearpc with a bootloader and very basic stripped down gnu/linux system, or even pearpc with its own kernel acting simply as a Hardware Abstraction Layer to boot you into OS X. You lose the cruft of having it run on a full operating system and would hopefully improve speed .

Too bad Transmeta doesn't have PPC firmware

[CustomDesigned]

Maybe there would be a market for a low power device that runs Win32, Linux, or MacOSX.

We used to have IBM 51x0 desktops. These were like Transmeta - they had a RISC CPU with a VM (CPU emulator) in ROM. There were two VMs available: System 360 (for running the System 360 APL interpreter) and System 36 (for running the System 36 Basic interpreter). There was a front panel switch to select the CPU emulation. Yes, like Transmeta, running the interpreter on top of the CPU emulator was fast enough to be very useful.

So, I am imagining a notebook with a front panel switch for i686/G4.

Re:OS X Panther Here

[ReadParse]

That's a good point. If there was a story today about a fantastic new computer that does in 10 minutes what previously took 7 hours, and it only costs a couple of thousand dollars, we would all be waiting in line to get it.

Don't get me wrong... I'm a cheap bastard also. But it's funny that people would actually go through this process. Emulation of x86 on a PPC makes more sense than the other way around, because if you were running both a Mac and Windows, you would certainly want to run the Mac as the HOST and the Windows as the DOG. The main reason emulation is needed is because there are hoards of Windows-only programs that Mac and Unix users have to use. Not so in reverse.

I appreciate the whole "because it's there" thing, but I don't think I'll be rushing to install OS X on an x86 machine when I have a perfectly good PowerBook (mine) and dual G5 (my employer's).

RP

Installs, but do apps run?

[sakusha]

Installs are easy, you're just copying files. But do apps run? The only reports I've read indicate that every app crashes immediately on launch, taking down the OS with it. Even clicking on the Dock causes a crash. This is not a successful install.

Apple (undoubtedly) HAS ported OS X to x86...

[neurocutie]

If only Apple would port the thing themselves.

There is no doubt that Apple already has OS X ported to the x86 architecture. What there is of OS X that isn't Darwin or FreeBSD (both run on x86) is largely based on the NextStep and NextStep was ported to the x86. So there is no doubt that OS X is around internally on x86. If ever Apple decides to give up on PPCs (not inconceivable (insert all old arguments about the difficulties of competing with x86/Intel/$10B chip foundries/etc)), then it must be ready with an OS X for the x86, so you know that Apple has x86 OS X internally just as a smart business precaution, to hedge its bets. But Apple is going to be extremely cautious about deciding to actually market and release x86 OS X... I think you can see why...

Re:OS X Panther Here

[arekusu]

It's worth pointing out that 10 minutes is for installing from optical drive, which is terribly slow.

If you run a lab, you install over gigabit ethernet via netboot, and your complete nuke&install happens in about THREE MINUTES, no joke.

Hey cool :-)

[ta bu shi da yu]

If they can make it work on non-32 bit architectures, then I can get OS X running on my old PowerPC that Apple have discontinued support for!

Re:Bye Bye Mac Hardware

[wvitXpert]

I was actually refering mainly to the case of the computer, not the internal components. I don't think that there can be much defference in the quality of the internal components, at least after you reach a certain (normal) level of quality. I just can't stand the toy like look and feel of practicly all other computers. I would like to know who decided that all PCs had to be made flimsy and cheap. As to the logic board failures, I'll still take my chances with Apple. I'd rather take the (slight) chance of getting a lemon Apple than being guaranteed of getting a crappy PC.