Haiku PowerPC compatibility

Oh I didn’t look at the NetBSD stuff, but yeah, I suppose I could use that instead of trying to fix ld. It was more of an exercise. I managed to generate a valid file although it wouldn’t load because it’s missing some stuff still.

Is there worked (semi-worked) files of Haiku PPC?

Anything that exists is in git or in the review queue… I don’t think PPC has seen much active work as of late. I think PPC and Sparc64 which Pulkomandy is working on a bit are roughly at similar stages neither boots to desktop yet.

Actually I do have some local branches here but only small attempts so not much.

1 Like

Still enough to make me technically wrong :wink:

1 Like

I really enjoy alternative architectures. I had a realization with PowerPC though that the hardware is pretty limited to:

  • Old Apple devices which aren’t wearing well (find me G5 iMac or iBook, and it’ll likely need its GPU reflowed/reballed to fix the video and the capacitors replaced)
  • Non-Apple PowerPC devices are pretty specialized and hard to locate.

The reasons above are why i transitioned to playing with arm,arm64 (greatly available) and riscv64 (cool factor). I see more value in @PulkoMandy 's SPARC64 work over PowerPC since SPARC hardware, (while old) is still holding its functionality well. (find me a SparcStation which doesn’t still grind away happily)

I make the occasional commit to PowerPC to try and keep our port building, but honestly unless someone picks up work on it, death is likely.


Yeah however pulkomandys port may not work on any sparcstation since those are all 32bit and not his particular interest…granted it should not be terribly hard to rectify that LOL. I bet Haiku would run quite reasonably on a quad 142Mhz hypersparc with 1GB ram (hyperstation 30 board). Or a sun4d…8x85Mhz 2GB.

Power hardware isnt that hard to get ahold of… like the Talos board etc…and its modern. There is also that open source PPC laptop being worked on…

If Be had focused on x86 instead of AT&T and PPC from the start they might still be around. A dual Pentium Pro BeBox would have been possible near the R3 release. Hindsight is 20/20 but Be miscalculated thinking CHRP would take off and that they could get around competing with Microsoft.

If Haiku were to focus on an architecture other than x86_64 it should be ARM, not PPC. I have a feeling the industry entire will be moving to ARM sooner than later leaving x86_64 behind and we will probably all be worse off for it because the x86_64 has been the closest thing we’ve ever had to an open hardware platform.


Well yes PPC isn’t as widespread as it used to be. The target is more of a sentimental value these days, but it’s also an incentive to optimize because those old machines have a low max-RAM point :sweat_smile:


Apple PowerMac G4 and G5 are plentiful.There are supporters for customized boards for FOSS development as well.

1 Like

Most G4 & G5s are preempted by MorphOS users :smiley:

those talos things cost like they are made of gold, where as they are not. too much overpriced. as of that laptop, this page leaves the impression, it’s a kindergarten:

wow. intrigued. :smiley: who would stay in doubts after such convincing arguments? :smiley: “was designed with more features”. :smiley: spherical features in vacuum. not serious.

Well it’s true that starting something 10 years later gives the possibility of not making the same mistakes and adding features correctly instead of shoehorning them (like, MMX reusing FPU regs on x86 forcing you to save them…). Not to say that it ends up as a competitive advantage, nor that competition is always fair (spoiler: it never is, else we’d all be running BeOS on BeBoxen nowadays :smiley:)…

I was just pointing out that they are not unobtainium neither are ultra expensive… neither does running on Power require that you buy vintage hardware.

Talos is comparable to similar PC hardware.

And last I checked the OpenPower laptop wasn’t targeting being a gold brick either.

If you primary goal is non x86 or having open firmware they are good targets… clearly not good targets if you are optimizing for value.

Everyone say that Be Inc. had to focus on x86. But actually they got the best tecnology for their time(power), acting with tecnical approach, then they choose the bigger market(x86), acting with a marketing appproach. It’s a pity that Sun opened sparc in 2005 and not earlier. An open source cpu… Afaik now sparc is abandoned by his stepmother Oracle, and the only company carrying on sparc processors is Fujitsu. I’ve always seen sparc a viable third way (under sun an ethical approach) specially after V9 for desktop use. But these are words of a dreamer, not a designer, nor a developer, neither a electronic engineer.

1 Like

It took me long time to realize, the destop segment was just a side-gig for Be Inc. I dont think they was ever serious about that.

I don’t think it was that it was a side gig, I think that it became a side gig too quickly.

tl;dr - Be Inc picked the wrong processor when they moved to PowerPC and it killed their hardware and desktop business. They then repeated the same mistake with the Internet Appliance business by sticking with Intel.

They were up against Microsoft in the desktop market and Microsoft of the 90’s ate little companies with niche OS for breakfast. They were blocked from the OEM market (and Be Inc got money for it after the company closed operations, save the one legal council that remained there to wind up their affairs - not that it helped them.)

The issue was this - they wanted to be like Next and Apple, but hardware is a terrible business to get in to without infrastructure. Look at what happened to Next’s hardware business - and Jobs invested in factories to produce the hardware.

They went with the Hobbit processor, and AT&T killed it and they fell in to using PowerPC rather than Intel - which actually turned out to be the death of the entire company. This key decision set them back years and effectively killed they entire business. If they had chosen Intel, they might still be around now, because Intel PC’s and hardware was all available as off the shelf - all they needed was a board with 2 processors, and getting that made is a lot easier IMO than designing an entire custom board.The rest could be PCI and ISA cards.

So the rest goes like this…they designed and manufactured the Bebox. But it was hard. Margins were tight. The business was not viable. Someone had the idea to port to other PowerPC machines, and that seemed sane given the investment in PowerPC. Apple was courting third parties for their doomed clone program. Apple was failing hard. They needed a new OS, and Be knew this. They started to bank on selling the OS to Apple, as this seemed like it would work - as heck, the OS already ran on the hardware! But that failed because Jobs was a better prospect, and Next was more established. And as a “thank you” present from Jobs and Apple, the rug got pulled again when they killed the clone program on Jobs return to Apple.

They could have partnered with Power Computing or someone like that to produce a new BeBox design, but that tied them in to what was effectively another hardware program - and I feel that the BeBox burned them so much that they decided that moving architecture would be safer.

They rushed R3 to Intel, and it more or less worked. But it was another dead end because the architecture they were using was based on PE format exes and that made the tooling horrible to work with and much more complicated. So they moved to ELF with R4 Intel. That burned a lot of early apps because there was no way to run anything from R3 on R4, yet on PowerPC you can still run stuff from PR1 and PR2 on R5.03.

They then went through the wilderness years of trying to make desktop sell. But that is not easy when you have the likes of Microsoft breathing down your neck constantly, trying to make you go away.

So, someone decided “let’s pivot” and so the great “Focus Shift” happened and they started the Internet Appliance venture… 10 - 15 years too soon. If they did it in 2010, if might have worked. But 2000 was a terrible time to be trying to make a small portable internet tablet device. And, with great irony, sticking with Intel basically killed them, because the Intel processors were really terrible at battery life. If they had used ARM or MIPS or something low powered, they would have been a lot better off.


It makes sense!

I guess from reading all the history, I see the sudden end of the clone program differently as a fan of both Apple and Be… afaik, the cloning was stopped by Jobs in the interests of creating a very tight focus and to save the company – not to “pull out the rug” from Be on purpose imho – although it did definitely have the end goal of “deep-sixing” the Mac clone companies. Still… I almost wish Apple could have been the ones to buy Be instead of Palm, revive the BeBox (and allowed Be to function as a partly independent subsidiary).

But the other sad thing is… Palm was brilliant imho (heck, I’m a Palm fan too) but had bad management (i.e. the Foleo looked really cool as a foray into the netbook space, and then it was undermined as a ‘mobile companion’ and scrapped; the LifeDrive exHD was a great idea but badly executed and the microdrives had problems, the Treo/Centro messed around with Windows on a few devices instead of having one solid product line, Cobalt failed, the webOS-powered Pre was way ahead but launched too late, and so on…) Saying all that because… I think Palm would’ve had the right idea. Imagine BeOS on the Foleo and a line of professional netbooks and that could sync to Palm’s handhelds/pilots seamlessly with their touchstone charger and Bluetooth – if Palm would’ve stayed alive, that would have been… awesomeness! :sunglasses:

Totally agree… I too think Be was way ahead of their time with the Internet Appliance idea – they could see it, but the technology to make devices thin and desirable was far from being there… and so it was a huge fail.

@apgreimann when Jobs killed the hardware clone business, Be lost access to all the documentation about the Apple devices being manufactured. It was at that point they stopped adding new PowerPC devices. I can even tell you the last models that were supported - the first revision of the 8600 and 9600. Apple changed the design of the motherboard on the last version of the 8600 and 9600 (the fastest revision) and made it incompatible, and from that point the not one further PowerPC model was supported that was not already based on an existing Apple reference design. No G3’s ever got any official support (I believe expansion cards worked in older Macs, and I heard very early Beige G3’s worked if you removed the personality card, though mine certainly didn’t.) and not one New World Mac (anything from the original iMac onwards) was ever supported. No version of MacOS past 8.6 ever booted BeOS also. So basically, at that point, and given PReP and CHRP never happened, they put all PowerPC support on the back burner.