-
Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.
New OS for PowerPC Macs
- Thread starter narke
- Start date
nglevin
Well-known member
Regarding GUIs, the last time I ran FreeBSD I was rocking WindowMaker.
It doesn't have anything to do with GNUstep now, but it has all of the NeXT style. And it's been getting regular updates recently.
It doesn't have anything to do with GNUstep now, but it has all of the NeXT style. And it's been getting regular updates recently.
commodorejohn
Well-known member
Themeable is good, but consistent underlying behavior is more important. That's where Linux really trips up, since there's approximately 47 different UI toolkits and 11,000 deviations from even those meager standards.
gsteemso
Well-known member
I agree 1000%. One of the few areas where OS 9 is still arguably superior to OS X is in UI consistency. Even with additions like the Windowshade functionality, classic Macs seem more predictable IMO. Later versions of OS X seem far more concerned with stylistic issues than usability ones, to the occasional detriment of consistency, and I agree with that one columnist at Ars Technica about all the many flaws associated with abandoning a spatially consistent UI model for the Finder. Of course, that can of worms is only tangentially related to the topic at hand, so I will stop there.
bse5150
Well-known member
While your idea is admirable, I think the PPC community would be better served with new software. Have you considered just porting some modern open source applications to the Mac?
No because I also want to support some wireless cards, support of other features like journaled file systems, preemptive multitasking, IPv6...
Just to be clear about this point, since it seemed to spur a pretty detailed discussion: the joke I was making by bringing up TempleOS (and comparing it to a variant of the AmigaOS) solely involved the "OS as a religion" aspect of it, I by no means was trying to make fun of the author's condition. That article I linked to is actually a pretty interesting read in how it points out some parts of TempleOS are actually surprisingly clever and sophisticated; the man is certainly not stupid, he just... isn't quite on the right rails.I feel sorry for Terry. Obviously a brilliant mind if it weren't for the schizophrenia.
----------------------
Anyway. Back to the original topic... regarding the pros and cons of making "just another UNIX-like OS", the unavoidable fact is that an OS with no applications to run doesn't really do anyone any good. HelenOS goes out of its way to *not* be UNIX-like and the documentation (such as it is) is full of gotchays and limitations regarding its POSIX compatibility. This OS is itself being positioned as essentially a PowerPC-specific *fork* of HelenOS and therefore unlikely to attract more than... realistically, what, a few hundred users? so it's highly unlikely developers are going to spend a lot of time considering it as a build target, which means that any "mainstream" software anyone wants is going to have to be laboriously ported over along with any necessary foundational libraries. I guess none of that really matters if ultimately your goal is to make the "OS" a complete suit of minimal applications and not really a target for third party development, but... well, I guess we'll see if people are willing to pay $40,000 for that.
(My personal favorite "tiny" operating system that doesn't really try to be "standard" in any way is probably KolibriOS. Of course, being written entirely in assembly language makes it somewhat nontrivial to port.)
If the whole motivation really just revolves around not wanting to make "Another UNIX" at all costs I wonder if starting with, I dunno, the Haiku or Syllable codebases might be a better place to start. Both of those at least have a small suite of GUI desktop applications and foundation libraries already ported.
I dunno, maybe I'm overestimating how hard it'll be to port things to whatever HelenOS uses for a GUI. Good luck in any case.
gsteemso
Well-known member
commodorejohn
Well-known member
On the other hand, practically nobody except OS geeks runs hobbyist OSes anyway, so it's not like you're alienating a huge potential userbase by not having a wide variety of preexisting applications, and if you go with Yet Another Freenix you're possibly losing what potential interest you might have had to begin with by having nothing much to set your project apart from any of the others.
Thank you.
I don't use Mac OS X, its too slow for iMac G3 and doesn't really run on Twentieth Anniversary Macintosh. Linux runs but there is an overhead at least for Xorg.
I agree.
Last edited by a moderator:
It's actually relatively common knowledge that Jobs himself would have preferred to stick with PowerPC hardware. Steve Jobs was, for all the growing he'd done at NeXT, still fairly capricious and was heavily susceptible to the belief that anything "not invented here" was a bad idea.
The problem is that in the real world, nobody had a good POWER or PowerPC road-map for what Apple needed to do, which was portable computers on one end of things, and computers that were "actually fast" on the other end of things. In 2005, Apple would still have found itself "competing" with $10,000 RISC UNIX workstations (the ones that would all die off in the next year or so, but hadn't quite yet), and so part of keeping a product like the Power Macintosh G5 relevant would have involved not slotting it in with the machines in the "five digit price tag" category. (Think: anything worth buying from Sun, HP(DEC/Compaq), IBM, SGI et al.) Unfortunately, part of doing that really meant using parts that just weren't actually as fast and good as what IBM was using in their own systems.
In terms of Mac OS X requirements -- (2000-2005) Part of the problem is that PowerPC was evolving extremely rapidly at that point. It had to in order ot keep up with Intel. Pentium 4s, for as terrible as they were, often massively out-performed G4s, and a good late-gen Pentium 4 on a 900-series chipset with DDR2 RAM was a force to be reckoned with, in terms of raw performance, and even moreso, in terms of performance per dollar.
With the rise of the web, and with information about Intel's road-map in the public eye, I'd like to think Apple had the foresight to know that PowerPC wasn't going to get them what they needed. this wasn't the '80s any more and you can't sell the same CPU from 1984 to 1990 and expect people not to look at you funny.
And, Apple was desperate. They weren't quite at the point of releasing a different-looking Power Macintosh with literally identical guts as the last generation (complete with an introduction video!) but I think Apple knew they were behind. As the 2000s wore on, the web started to be a thing that really drove performance improvements. Without that, Macs could probably have plodded on as productivity and media machines, with the default narrative being that you buy a Mac not because it's fast, but because it gets your work done well or because they're reliable.
Although the exact cut-offs were often arbitrary, I think that they happened because Apple needed a convenient way for people to self-identify whether or not their machine would run a program or an OS. An oft-cited example is Mac OS X 10.3 or 10.4, which "required" a firewire port. It did not actually need a firewire port to run, but the firewire port designation was convenient because most G3 Macs that didn't have firewire also lacked a powerful enough GPU, a big enough disk, and enough RAM to run the software well. And so, one or two machine that probably could have done well got cut off with a whole group of them that just had no chance.
In terms of dumping PowerPC Macs once the Intel switch was made? They got a few years. It wasn't until Mac OS X 10.7.0 was released that the last of the PPC Macs stopped getting software patches from Apple, and I think most developers had already given up of their own accord by then, not at Apple's behest. Microsoft, for example, has long had a habit of making Office available only for the most recent one or two versions of Mac OS X. Adobe has been the same, and I think part of that has been that classically, Office and Creative Suite had very long life-cycles. Adobe's release cycle quickened a lot during the 2000s and early 2010s but Microsoft's doesn't really show that much indication of speeding up.
G5s may not have "a million" cores, but mainstream PCs don't either. The only things with more than four cores at this point are enthusiast platforms, workstations, and mobile phone platforms that have "eight cores" (whereby it's two quad-core designs glued together, one designed for performance, and one designed for low power usage.)
Ultimately, the problem with calling the G5 a really useful computer is that for $700 you can pick up something with a 5-watt dual core processor running at less than a gigahert (base frequency) that will happily trounce the G5 at everything. You might need to tdpUp the chip or move to a 15-watt chip to actually be twice as fast as a quad, but it's not going to be hard or expensive.
The G5 served a purpose in the 2000s. So did the Octane2 and so did the C8000 and so did the Ultra45. Suggesting that it's a suitable computer for a casual computer user today, however, just isn't in the cards. "Barely able to keep up with modern netbooks," is probably a better way to talk about G5 performance (let alone G4 performance) and the most relevance it has is either doing whatever it did in 2005, or working for a hobbyist.
At the end of the day, it's somewhat disingenuous to talk about the very last generation of Power Macintosh G5s as though they're at all representative of anything else Apple had built up to that point. Physical reliability aside, the Quad is over twice as fast as every other G5, including the dual 2.7GHz model. The other models are basically in line with normal G5 performance, close to the top of the heap, but they didn't exactly set the world on fire. Normal G5 performance is pretty close to what you'd see out of that 2.0GHz dual-core G5. Even more systems were dual 1.8GHz models. And so, thinking about comparisons to that system are probably better.
So suggesting that just because the Quad (if you equip it with the best of everything) is well suited to a particular task doesn't mean any other G5 (especially those that can't run 16 gigs of RAM) are suited to that task. They might be able to do it, but it won't necessarily be fun.
The idea of developing a "new" operating system for old PPC Macs is pretty interesting. It's obvious that PowerPC distributions of Linux and BSD already exist, but those are problematic, for reasons. (Mostly because the reason to buy a Mac was never because of its hardware, but because of the software, and the integration between the software and the hardware that Apple can build.)
It's interesting to see hobby operating systems develop, but I would honestly be more interested in it if it were an OS for a Tegra development board, or for the Raspberry Pi computer. Those things are going to be more inclusionary for the development of a new OS anyway (because it's easier and cheaper to get a Pi than most PPC Macs), those things are quieter and lower-power to run, and I would actually bet money that the new Jetson TX1 development kit is faster than a G5 Quad. (Though, you'll pay through the nose for a Jetson TX1. No longer is nVidia doing the cheeky "$1/CUDA Core!" thing they did with TK1, which was $192.)
All of that said, the Pi is probably the best place if you want a non-Intel platform to develop a new hobbyist OS. There's modern POWER hardware from IBM and Tyan, but that, which actually cheaper than RISC UNIX workstations of yore, is still not exactly something you just sort of buy on the debit card one afternoon because you felt like it might be neat.
The Apple IIgs as well. It's interesting to see what happened to the platforms that died in the '80s and '90s. Hobbyists have recently released GS/OS 6.0.3 for example, and I've seen interesting stuff about Atari TOS (and a replimentation thereof) on coldfire hardware, and the plight of the Amiga user is well known at this point.
I think part of why none of that ever happened for the Mac is because Apple and the Mac each continued existing. If you go purely by names (which you have to to believe that the current new Amigas are really related to classic 68k Amigas) then you understand that a MacBook Air is the logical ultimate successor to the Mac Portable.
PowerPC stuff is even common in some networking equipment at this point.
The question is whether or not developing a modern OS is really relevant for users of PPC Macs. Most people who get into the "Mac" do so because they wanted a Mac, not specifically because they wanted a PowerPC computer, or because (somehow) they think that any of the architectures Apple was building were any good. (Unless you judge goodness only by what came before, and you also conveniently forget that anything with a G3 ever existed.)
Linux and BSD for PowerPC Macs are therefore problematic because they say "well, the hardware's still relevant, but you have to give up the software you were running on it." The SGI IRIX community has essentially dismissed this out of hand, in part because Linux/BSD support for SGI's MIPS hardware is completely abysmal, and in part because the whole reason you used IRIX (ever) was because of its applications, not because you actually thought IRIX was a really good UNIX system. (It was passable, but security was worse than in everything else. You were probably there for the graphics cards or for the scale-up high performance systems SGI built with what they got from Cray.)
Reimplementations of an existing OS are probably more interesting than Linux/BSD on PPC Macs. If there was some modern operating system that ran on faster powerPC hardware that claimed to support running Mac OS 7/8/9 applications, for example, I'd be pretty interested. I'm also a bit more interested in the implications of a hobby/demo OS such as this, but I'll be honest, as somebody else has already pointed out, this OS already runs better on x86 hardware that's faster, more reliable, easier to come by, and more convenient to use in most cases. (Especially any case where performance is important.)
The implications of MorphOS on Mac/PPC hardware are pretty similarly interesting, but it looks like the Amiga/MorphOS people are really moving in the direction of new boards with QuorIQ CPUs, and things like the X1000 -- Probably in an attempt to get something a little more sustainable for themselves, something a little faster and less brain-dead, design-wise, more flexible, and closer to the original Amiga design goals, which as I've read, favored the GPU heavily over the CPU.
Ultimately, I think it depends on whether the goal is "make a Mac more useful" or "run a more useful OS on this hardware" -- each of these things has different implications and is going to appeal to a different set of people.
This project essentially aims to build an OS more useful than Classic Mac OS, and more... (secure? updated? relevant? interesting?) than Mac OS X. It's going to be hard in a lot of ways.
Hardware support is going to be insane, for example. Because the list of "supported" Macs is essentially anything with PCI or newer (so, you're talking, 7200/75 with its blisteringly fast Pentium-crushing 75MHz 601 CPU is ostensibly supported) you're talking about building something that scales well from a single 75MHz 601 to a quad 2.5GHz G5. I'm not always super supportive of the Quad, but even I have to admit that it is faster than a 7200. Not only are you talking about a big difference in processor speed, but a 7200 with 8 megs of RAM isn't impossible to find. Building one OS that'll run in both 8 (or heck, 64) megs of RAM and 16 gigs isn't going to be an easy task. I suspect most of this is why you don't hear about Linux for oldworld at this point. Even heavily upgraded oldworld machines (and, they need to be heavily upgraded) will still not be fast as a good dual G4 or most G5s.
The other thing is that because you're ultimately proposing either building a new scene, or converting an existing scene over to a new OS, there's a lot of software work to be done.
One thing that would be really interesting ot know is of the people who are still using late PPC machines, what are they doing on them? Is it really just a bunch of people using tenfourfox to talk about how terrible Apple is on here and macrumors, and visit blogger sites? Or are people still using period apps like final cut, aperture, the whole creative suite, and even higher end things like Maya, Mathematica, and whatever?
I have my own suspicions.
On the other hand, because of my suspicions, the people who are using Macs for productivity with those kinds of programs have probably long since moved on to a more modern Mac or Windows computer, leaving the remaining Mac/PPC user base susceptible to suggestions of a switch to a new OS on their existing hardware.
though, you still need to consider the fact that people who are on Macs are there because they're Mac-Like. In ~1986-1990, the motivations for buying an Apple II, Mac, ST, or Amiga were all totally different. Most Mac users were there because they wanted to get away from some of the really technical aspects of daily life on an Apple II or Amiga.
Selling Mac users on an experience really designed for Apple II, Amiga, or ST users isn't going to be easy. Perhaps this is one of the more interesting things about the early days of Mac OS X -- Apple was really building systems capable of a few different experiences, and the $10,000 UNIX workstation market was definitely interested in Apple, because they were building competing machines for a lot cheaper. At the time, Mac users were legitimately worried that it was going to spell the end of their platform as they'd known it -- an easy and predictable graphical environment for people who wanted to get work done.
Which really old stuff? A hobbyist group recently released a new version of GS/OS based on the Apple IIgs System 6 source code, and as of yet, Apple hasn't sued.
The iPhone/Samsung suits were started when those were still relevant, and are based on product famlies and brand names ("iPhone" and "Galaxy") that are still relevant to both Apple and Samsung, if that's what you're talking about.
I also don't believe I've heard about GNOME or MATE being sued, and that's configurable to look and work very similarly to Mac OS (both Classic and X.)
I doubt Apple's gonna sue if some new OS or window manager comes out and it looks sort of like Classic Mac OS.
While I don't believe letting Mac OS X on the Internet or even on Internet-connected LANs is a good idea, I actually think this is a much better idea than a whole new OS. See my comments about "Mac-Like" above.
The reason people buy Mac hardware, new or used, is because they want Apple's integration of OS and hardware.
Anybody who wanted to run a hobbyist OS would probably buy a Pi, and anybody who wanted to run Linux/PPC would be best served by a box from IBM or Tyan. (Or perhaps some of the newest boards that the Amiga crowd is building.)
It's interesting to see hobby operating systems develop, but I would honestly be more interested in it if it were an OS for a Tegra development board, or for the Raspberry Pi computer. Those things are going to be more inclusionary for the development of a new OS anyway (because it's easier and cheaper to get a Pi than most PPC Macs), those things are quieter and lower-power to run, and I would actually bet money that the new Jetson TX1 development kit is faster than a G5 Quad. (Though, you'll pay through the nose for a Jetson TX1. No longer is nVidia doing the cheeky "$1/CUDA Core!" thing they did with TK1, which was $192.)
All of that said, the Pi is probably the best place if you want a non-Intel platform to develop a new hobbyist OS. There's modern POWER hardware from IBM and Tyan, but that, which actually cheaper than RISC UNIX workstations of yore, is still not exactly something you just sort of buy on the debit card one afternoon because you felt like it might be neat.
The Apple IIgs as well. It's interesting to see what happened to the platforms that died in the '80s and '90s. Hobbyists have recently released GS/OS 6.0.3 for example, and I've seen interesting stuff about Atari TOS (and a replimentation thereof) on coldfire hardware, and the plight of the Amiga user is well known at this point.
I think part of why none of that ever happened for the Mac is because Apple and the Mac each continued existing. If you go purely by names (which you have to to believe that the current new Amigas are really related to classic 68k Amigas) then you understand that a MacBook Air is the logical ultimate successor to the Mac Portable.
PowerPC stuff is even common in some networking equipment at this point.
The question is whether or not developing a modern OS is really relevant for users of PPC Macs. Most people who get into the "Mac" do so because they wanted a Mac, not specifically because they wanted a PowerPC computer, or because (somehow) they think that any of the architectures Apple was building were any good. (Unless you judge goodness only by what came before, and you also conveniently forget that anything with a G3 ever existed.)
Linux and BSD for PowerPC Macs are therefore problematic because they say "well, the hardware's still relevant, but you have to give up the software you were running on it." The SGI IRIX community has essentially dismissed this out of hand, in part because Linux/BSD support for SGI's MIPS hardware is completely abysmal, and in part because the whole reason you used IRIX (ever) was because of its applications, not because you actually thought IRIX was a really good UNIX system. (It was passable, but security was worse than in everything else. You were probably there for the graphics cards or for the scale-up high performance systems SGI built with what they got from Cray.)
Reimplementations of an existing OS are probably more interesting than Linux/BSD on PPC Macs. If there was some modern operating system that ran on faster powerPC hardware that claimed to support running Mac OS 7/8/9 applications, for example, I'd be pretty interested. I'm also a bit more interested in the implications of a hobby/demo OS such as this, but I'll be honest, as somebody else has already pointed out, this OS already runs better on x86 hardware that's faster, more reliable, easier to come by, and more convenient to use in most cases. (Especially any case where performance is important.)
The implications of MorphOS on Mac/PPC hardware are pretty similarly interesting, but it looks like the Amiga/MorphOS people are really moving in the direction of new boards with QuorIQ CPUs, and things like the X1000 -- Probably in an attempt to get something a little more sustainable for themselves, something a little faster and less brain-dead, design-wise, more flexible, and closer to the original Amiga design goals, which as I've read, favored the GPU heavily over the CPU.
Ultimately, I think it depends on whether the goal is "make a Mac more useful" or "run a more useful OS on this hardware" -- each of these things has different implications and is going to appeal to a different set of people.
This project essentially aims to build an OS more useful than Classic Mac OS, and more... (secure? updated? relevant? interesting?) than Mac OS X. It's going to be hard in a lot of ways.
Hardware support is going to be insane, for example. Because the list of "supported" Macs is essentially anything with PCI or newer (so, you're talking, 7200/75 with its blisteringly fast Pentium-crushing 75MHz 601 CPU is ostensibly supported) you're talking about building something that scales well from a single 75MHz 601 to a quad 2.5GHz G5. I'm not always super supportive of the Quad, but even I have to admit that it is faster than a 7200. Not only are you talking about a big difference in processor speed, but a 7200 with 8 megs of RAM isn't impossible to find. Building one OS that'll run in both 8 (or heck, 64) megs of RAM and 16 gigs isn't going to be an easy task. I suspect most of this is why you don't hear about Linux for oldworld at this point. Even heavily upgraded oldworld machines (and, they need to be heavily upgraded) will still not be fast as a good dual G4 or most G5s.
The other thing is that because you're ultimately proposing either building a new scene, or converting an existing scene over to a new OS, there's a lot of software work to be done.
One thing that would be really interesting ot know is of the people who are still using late PPC machines, what are they doing on them? Is it really just a bunch of people using tenfourfox to talk about how terrible Apple is on here and macrumors, and visit blogger sites? Or are people still using period apps like final cut, aperture, the whole creative suite, and even higher end things like Maya, Mathematica, and whatever?
I have my own suspicions.
On the other hand, because of my suspicions, the people who are using Macs for productivity with those kinds of programs have probably long since moved on to a more modern Mac or Windows computer, leaving the remaining Mac/PPC user base susceptible to suggestions of a switch to a new OS on their existing hardware.
though, you still need to consider the fact that people who are on Macs are there because they're Mac-Like. In ~1986-1990, the motivations for buying an Apple II, Mac, ST, or Amiga were all totally different. Most Mac users were there because they wanted to get away from some of the really technical aspects of daily life on an Apple II or Amiga.
Selling Mac users on an experience really designed for Apple II, Amiga, or ST users isn't going to be easy. Perhaps this is one of the more interesting things about the early days of Mac OS X -- Apple was really building systems capable of a few different experiences, and the $10,000 UNIX workstation market was definitely interested in Apple, because they were building competing machines for a lot cheaper. At the time, Mac users were legitimately worried that it was going to spell the end of their platform as they'd known it -- an easy and predictable graphical environment for people who wanted to get work done.
Which really old stuff? A hobbyist group recently released a new version of GS/OS based on the Apple IIgs System 6 source code, and as of yet, Apple hasn't sued.
The iPhone/Samsung suits were started when those were still relevant, and are based on product famlies and brand names ("iPhone" and "Galaxy") that are still relevant to both Apple and Samsung, if that's what you're talking about.
I also don't believe I've heard about GNOME or MATE being sued, and that's configurable to look and work very similarly to Mac OS (both Classic and X.)
I doubt Apple's gonna sue if some new OS or window manager comes out and it looks sort of like Classic Mac OS.
While I don't believe letting Mac OS X on the Internet or even on Internet-connected LANs is a good idea, I actually think this is a much better idea than a whole new OS. See my comments about "Mac-Like" above.
The reason people buy Mac hardware, new or used, is because they want Apple's integration of OS and hardware.
Anybody who wanted to run a hobbyist OS would probably buy a Pi, and anybody who wanted to run Linux/PPC would be best served by a box from IBM or Tyan. (Or perhaps some of the newest boards that the Amiga crowd is building.)
TL;DR
It's an interesting idea, but trying to sell it to Mac users specifically is probably fairly misguided.
Ultimately, it'll probably be easier to develop it for the Pi, you're likely to get more people interested in donating to that cause, and the OS and its goals honestly sound better suited to hardware like the Pi, or even high end developer hardware like one of the nVidia development boards (Kayla, Jetson TK1, Jetson TX1) than to Mac hardware.
It's an interesting idea, but trying to sell it to Mac users specifically is probably fairly misguided.
Ultimately, it'll probably be easier to develop it for the Pi, you're likely to get more people interested in donating to that cause, and the OS and its goals honestly sound better suited to hardware like the Pi, or even high end developer hardware like one of the nVidia development boards (Kayla, Jetson TK1, Jetson TX1) than to Mac hardware.
nglevin
Well-known member
I agree with this, but I suspect that there might be a better solution in between.
Freescale Semiconductor produces Power architecture development boards that could be sourced and replaced without having to rely on second hand hardware, and they have versions that run on ARM too. They're designed to be stacked with custom made LCD screens and they come with a Linux distro. I see no reason why you can't upload a custom OS to one of these to play around with before scaling it down to work with older Macs and the "old world"/"new world" firmware.
(Bearing in mind that I do very little OS development myself, so there may be significant roadblocks that I'm overlooking. If NetBSD can be made to run on a Mac Quadra, though, this doesn't seem like so much of a leap.)
EDIT: And of course, Freescale still sells a version of CodeWarrior for these PowerPC boards. It's about as close as you can get to PPC Mac development with today's hardware and software.
Freescale Semiconductor produces Power architecture development boards that could be sourced and replaced without having to rely on second hand hardware, and they have versions that run on ARM too. They're designed to be stacked with custom made LCD screens and they come with a Linux distro. I see no reason why you can't upload a custom OS to one of these to play around with before scaling it down to work with older Macs and the "old world"/"new world" firmware.
(Bearing in mind that I do very little OS development myself, so there may be significant roadblocks that I'm overlooking. If NetBSD can be made to run on a Mac Quadra, though, this doesn't seem like so much of a leap.)
EDIT: And of course, Freescale still sells a version of CodeWarrior for these PowerPC boards. It's about as close as you can get to PPC Mac development with today's hardware and software.
Last edited by a moderator:
There are loads of PowerPC development boards out there, but I have yet to really se any good reason why one is better than an ARM or MISP development board, even a regular x86 board.
The Pi is like $35, for a quad-core CPU and a gig of RAM. It's going to be more flexible and have more resources than the cheapest PPC dev kits (the linked model is $225 for a dual-core CPU and 512 megs of RAM) and a lot cheaper than the more flexible dev kits.
There are more desktop/server focused boards available, but I guess the question is at that point, why bother with PowerPC at all? More people will benefit from and be able to contribute to a project centered around a $35 Pi than around a more expensive system, and the Pi will still outgun most PPC Macs.
The Pi is like $35, for a quad-core CPU and a gig of RAM. It's going to be more flexible and have more resources than the cheapest PPC dev kits (the linked model is $225 for a dual-core CPU and 512 megs of RAM) and a lot cheaper than the more flexible dev kits.
There are more desktop/server focused boards available, but I guess the question is at that point, why bother with PowerPC at all? More people will benefit from and be able to contribute to a project centered around a $35 Pi than around a more expensive system, and the Pi will still outgun most PPC Macs.
nglevin
Well-known member
It's a fair point, how is PowerPC competitive with ARM when even the former Motorola company's moving towards ARM chips. One PPC architect I knew who worked at Apple actually worked at ARM for a time, thought of their architecture as the closest thing to modern Mac silicon.
Personally, I absolutely hated working with the Pi because of a unique situation where the GPU is absolutely incapable of setting up more than one "surface" for rendering with OpenGL. Which means that most everything including Google Chrome and the default Raspbian desktop is piped through the CPU with MESA and X11. Which means that you'll have to overclock for anything resembling reasonable desktop performance, and it still feels pokey compared to a 68k Mac. Which has basically deadlocked efforts to update the Raspbian platform with Wayland or a similar, "modern" GUI platform for Linux development. All that despite the project's close ties to the GPU vendor, Broadcom. Too many politics about that project for not enough positive ROI.
IMHO, the GPU's a bit of a lost cause unless you're looking for something to just render some video or do some graphics demos. You'd be surprised at how many RasPi 3D demos are actually running on the CPU with MESA.
Getting back to your point Cory, I'd be happy to tinker with PPC as a weekend project if I had the time.
Maybe there wouldn't be much support for a project like that, but then, I question how many developers there really are who might be interested in an OS project outside of BYO Linux distro. There's a large pool of developers with a shallow understanding of the technology they work with, and only a small minority that really want to color outside the lines and dig deeper into truly strange areas of hardware and software. If there are friends willing to come along for the PPC OS ride, that's enough of a reason to start. And honestly, I'd take a long hard look into finding peers willing to keep the project going before really starting on the architecture.
The current state of software development's sad and pathetic enough that even some willingness to write a Linux kernel extension is more than enough to stand out from the pack. I don't see much present day demand in building a new operating system, but if that's what fires you up? Why not.
Similar reasoning could be used to ask, why build your own game engine. I tried, started, got quickly bored because nobody wanted to carry it on. But I never regretted trying.
Personally, I absolutely hated working with the Pi because of a unique situation where the GPU is absolutely incapable of setting up more than one "surface" for rendering with OpenGL. Which means that most everything including Google Chrome and the default Raspbian desktop is piped through the CPU with MESA and X11. Which means that you'll have to overclock for anything resembling reasonable desktop performance, and it still feels pokey compared to a 68k Mac. Which has basically deadlocked efforts to update the Raspbian platform with Wayland or a similar, "modern" GUI platform for Linux development. All that despite the project's close ties to the GPU vendor, Broadcom. Too many politics about that project for not enough positive ROI.
IMHO, the GPU's a bit of a lost cause unless you're looking for something to just render some video or do some graphics demos. You'd be surprised at how many RasPi 3D demos are actually running on the CPU with MESA.
Getting back to your point Cory, I'd be happy to tinker with PPC as a weekend project if I had the time.
Maybe there wouldn't be much support for a project like that, but then, I question how many developers there really are who might be interested in an OS project outside of BYO Linux distro. There's a large pool of developers with a shallow understanding of the technology they work with, and only a small minority that really want to color outside the lines and dig deeper into truly strange areas of hardware and software. If there are friends willing to come along for the PPC OS ride, that's enough of a reason to start. And honestly, I'd take a long hard look into finding peers willing to keep the project going before really starting on the architecture.
The current state of software development's sad and pathetic enough that even some willingness to write a Linux kernel extension is more than enough to stand out from the pack. I don't see much present day demand in building a new operating system, but if that's what fires you up? Why not.
Similar reasoning could be used to ask, why build your own game engine. I tried, started, got quickly bored because nobody wanted to carry it on. But I never regretted trying.
gsteemso
Well-known member
I, like many nuts-n-bolts computer/retrocomputer hobbyists, have a long list of projects I occasionally devote brain space to. Some of them are conceivably attainable; many more would theoretically be possible, but only if there were a few hundred of me and an unlimited budget. Among the latter, which I fondly think of as Impractical Pipe Dream Projects (IPDPs -- yes, I have so many as to have developed a nomenclature), is a ridiculous plan to reverse-engineer classic Mac OS and write an improved version "Like Mac OS 8 (and X) Should Have Been". The basic idea is to write a protected, preemptive, hardware-independent OS, able to run normal Mac software, that could take full advantage of later Mac CPU hardware like PPC (...and, reluctantly because it's already been done to death, x86-64) -- but to do it using natural extensions of classic Mac OS software technologies, rather than completely displacing everything with some alien beast based on NextStep and UNIX, like Apple actually did.
I am 100% certain that I have neither the skills nor the resources to actually get anywhere meaningful with such a project, but startlingly, it seems a good starting point in the form of the source code to System 7.1 has semi-recently been leaked onto the net. Naturally, a surprising amount of that code is written directly in 68k assembly language, but it's something.
That's where I would start with any OS development project for vintage Mac hardware -- a system made to run Mac software in a Mac-like way. As pointed out earlier in this thread, we have Macs as opposed to some other platform for (what we, at least, feel are) good reasons. It only makes sense to work with that. Such a plan would also take care of the "But there's no software for it, what's the point?" problem someone else brought up.
I am 100% certain that I have neither the skills nor the resources to actually get anywhere meaningful with such a project, but startlingly, it seems a good starting point in the form of the source code to System 7.1 has semi-recently been leaked onto the net. Naturally, a surprising amount of that code is written directly in 68k assembly language, but it's something.
That's where I would start with any OS development project for vintage Mac hardware -- a system made to run Mac software in a Mac-like way. As pointed out earlier in this thread, we have Macs as opposed to some other platform for (what we, at least, feel are) good reasons. It only makes sense to work with that. Such a plan would also take care of the "But there's no software for it, what's the point?" problem someone else brought up.
nglevin
Well-known member
A good first step for that would be to document how to build the source code, and what tools to use with that.
Really, for whatever OS project happens to be undertaken, figuring out what milestones to reach is absolutely crucial, and how to keep iterating fast through them without getting too concerned about optimizing X and Y as you pick up on inefficiencies here and there.
If there's anything I can contribute to this thread besides silly rambling about the status quo;
1) Find people who also want to undertake the project and can contribute
2) Figure out what deliverables you can produce on a short timeframe, and quickly iterate towards those goals.
That'll help you scale to something bigger, if you still want to move forward with this project once you start producing bits that you can show off and make (strange and unusual!) people excited about what's next.
Last edited by a moderator:
IPalindromeI
Well-known member
RISC OS and Plan 9 seem pretty popular on Pi, and I hear QNX has been ported.
Similar threads
