Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
Linux Software

Linux And The PowerPC Architecture 202

Linux is always a little bigger than you think. Every day, people are working on porting Linux to new platforms to achieve more with what they have. In the case of LinuxPPC and Terra Soft Solutions, they're working with community spirit and the PowerPC architecture to create insanely great solutions with a touch of Tux.

For those of you just joining us, here's a little background on the PowerPC architecture from the PowerPC FAQ, maintained by Gary Davenport and readily available at linuxppc.org.

[2.1] What is PowerPC?

PowerPC is an architecture which was jointly designed by Apple, IBM, and Motorola. The PowerPC architecture specifies an instruction set architecture (ISA), which allows anyone to design and fabricate compatible PowerPC processors.

The PowerPC architecture is derivative of the IBM POWER architecture, used in IBM's RS/6000 workstations.

[2.2] What is POWER? How does it relate to the PowerPC?

The POWER architecture was introduced by IBM for use in its RS/6000 systems in early 1990. Its name an acronym for "Power Optimization With Enhanced RISC", POWER was one of the first superscalar RISC designs. It was originally a multi-chip implementation, but diversifying workstation needs prompted work on a single-chip version.

As Apple came to IBM for its microprocessor needs, the need for a single-chip RISC processor became urgent, and the work done for the single-chip POWER derivative was polished and paired with Motorola's 88K processor bus. This was the PowerPC 601, which made its debut in the very first Power Macintoshes on March 14, 1994.

The POWER family proper advanced with the POWER2 in late 1994 and the POWER3 in late October, 1998.

The Macintosh stands on its own in the 'most-loved' personal computer category, and has won unbelievable customer loyalty through its unorthodox appearance and powerful interface. No strangers to the world of ideological zealotry, the Mac-faithful are bringing new power to the platform with the time-honored tradition of hardware hacking and GPL'ed software design. I recently spoke to Terra Soft Solutions CTO Dan Burcaw about what drives him to work on Linux for the PowerPC architecture.

"I love working on the latest Apple machines," Dan said. "Linux on a Power Macintosh G4 is just incredible. Besides Apple systems there are a variety of other cool PowerPC boxes out there like IBM RS/6000s and embedded boards. As a change, they are fun to work with too.

"Also, I really enjoy the community. The PowerPC Linux market is much, much smaller then the x86 market and it makes for more of a close community.

"Just to clarify, I don't work on LinuxPPC the product but I do work on the general PowerPC Linux effort. There has been a great deal of confusion between LinuxPPC.org (the home page for the Linux port to PowerPC) and LinuxPPC, Inc., the company that produces the LinuxPPC distribution.

"We tend to refer to the general port as PowerPC Linux and LinuxPPC refers only to the company."

Speaking of porting software, you'd be amazed how much of it compiles just fine on the PowerPC architecture. Since a great number of tools are available in source code form for Linux, porting it over to another platform isn't that difficult, and crosses the great divide between the 'IBM' and 'Macintosh' sections of the virtual software store. I also got to speak to Ani Joshi, a University of Maryland student that's also working hard to make more things work on PowerPC Linux.

"When I go about porting software, I first build it alongside with my x86 machines. I make sure there is no inline assembly which could cause problems, and if there is then I break out my x86 assembly manuals and start porting to straight PowerPC asm for those routines. Otherwise our glibc (thanks to our gcc/glibc man Franz Sirl) is very helpful and will assist me in any cross-platform code in the app. If there is some x86 non-asm code in there, it should not be a problem for us as we have very good userland libraries which allow a clean powerpc 'solution' to the problem. As for performance in LinuxPPC, I'd say people will be very happy with the G4's performance, especially once more people start enhancing software for the AltiVec unit. Our CPU's speak for themselves; the PPC platform is quite strong, in our humble opinion."

What about the artists and musicians who have called the Macintosh 'home' for so many years? Where do they fit into the new Linux landscape of Linux on the PowerPC architecture? Back to Dan Burcaw:

"Well, obviously programs like the Gimp are very important to these folks. Most of them right now will probably still keep MacOS around for Photoshop and other programs. I think we'll be able to better serve their needs as software comes over to the platform. This is a major challenge. Software companies need hard data to show that there is demand for this software. As I said, PowerPC Linux is actually a very small market and so it is hard for a lot of these companies to justify the ports (even if it is just a simple recompile)."

Dan's Black Lab Linux Project Lead, Troy Benjegerdes, adds:

"What this DOES help out a great deal on is in the science and engineering areas. The G4 processor is absolutely ideal for dealing with any kind of signal or image processing because of the AltiVec unit. Combine that with the flexibility of Linux, as well as Beowulf Clustering technology, and you have an environment for signal and image processing that very few other things can even come close to."

Porting end-user apps to Linux on another platform is great, but what about expanding the original project to include hypermachines? Troy's got game in the parallel computing department, and Black Lab Linux is specifically designed for it.

Troy says, "Recently, I've been working on porting Black Lab Linux to CSPI hardware. CSPI is a company that makes what could best be described as an 'embedded supercomputer'. They make PowerPC processor boards that are extremely compact; You can put 64 G4 processors in a single cabinet that takes up 9U on a 19" rack. If you fill an entire 7 foot tall 19 inch rack with CSPI systems, you will have over 250 G4 processors. Thats a 250 node Beowulf with a peak performance of up to 1 TeraFLOP that will fit in a closet!"

Hardware and software innovation is one thing, but to make money in this business, you've got to be able to support what you're doing. I spoke with Hollis Blanchard, who works in tech support for LinuxPPC, Inc., and I asked him about his most common LinuxPPC support problems.

"Booting is a nightmare. Open Firmware (in theory) is supposed to make life easier to boot different OS's, but I can only dream of PC BIOS. All the OF in the old Macs is horribly broken in one way or another (like on my machine, it can't drive my video, so there's no way to see what I'm typing). New machines are better, but there are so many obstacles and the hackers just don't have enough time.

"There's also a philosophical problem sometimes. Some of the people we talk to have heard all these wonderful things about Linux and are expecting it to be as easy and polished as the Mac OS. That's a hard standard to live up to. It's NOT easy - Don't believe the hype. Putting Linux on the desktop still very much depends on educating the desktop user. A lot of them have no interest in such education and rightly so."

Linux Ease-of-use issues certainly aren't indigenous to the PowerPC world, and it just goes to show that although these guys aren't working on x86, they have the same issues. I think Hollis sums it up perfectly.

"Mad props to everyone who's ever submitted a patch that fixed something for PPC. There are too many naidne-elttil x86-only folks out there... you have to remember the world is bigger than the US."

This story was inspired by and dedicated to James A. Irwin.

This discussion has been archived. No new comments can be posted.

Linux And The PowerPC Architecture

Comments Filter:
  • by Anonymous Coward

    PowerPC systems are great, but Linux on them is also slow. If it weren't for this, I would buy Macs for my linux systems, as I love the hardware. But the linux performance ain't there. I even ran nbench as a test, and my PII350 embarrassed my G3400 running RedHat 6.1 and LinuxPPC 2000 respectively.

    The compiler just doesn't optimize code well. TerraSoft has the Altivec support, but that would require a large effort to get everythign using it and I still haven't seen where I can get their compiler...
  • And this, unfortunately, really is the main problem with PPC and any OS--the lack of enough market support and alternative products. Sure, you can run a PPC LInux distro on a Mac, but it would be so much better if we had even one or two sources of PPC motherboards.
  • by Anonymous Coward
    A plea to Linux hackers everywhere.

    Abandon ship. Port over to NetBSD and stop dicking around with x86-only code.
  • by Anonymous Coward
    MacOS GUI is certainly easy, simple and clean, but it is not intuitive. No computer interface is ever intuitive, some just have have a less steep learning curve. Make no mistake -- easy to learn is NOT the same thing as intuitive. The MacOS GUI can be powerful, but the overall feeling of the system is really hampered by the primitive inner workings of the traditional MacOS (no pre-emptive multitasking). Even though it might not be such an issue to an end user, it actually slows down learning as well. True 'power using' is very difficult with the MacOS. So, the MacOS 10 will correct all that, but alas it is late, oh so late! Originally the MacOS 8 was supposed to have pre-emptive multitasking (the Copeland project). The current version number is 9.0.x, and still multitasking slows everything down to crawl on a G3 machine, especially if there are some kind of I/O activities. Needless to say, Linux on the same machine is considerably more responsive.
  • I never said that they were giving it away. That price includes Office Pro Academic for $99. (At any time it may rise to the standard $180 academic price, but that system still torches the Mac)
  • I was mainly speaking of price/performance here, not top-end availability. Yes, Apple's PB selections is somewhat limited. But if you're looking for a laptop so that you have something portable when you need it, the iBook is a great deal. If you need something in the Powerbook range, they also beat PC systems in price/performance. (At least anything from Dell/Compaq/Gateway/IBM/Toshiba, the 5 vendors we sell at the CStore.) IMHO getting a PIII/650 laptop with a 15.4" screen is insane - Get a much more basic laptop and a faster desktop, you'll still be spending less.
  • I sold computers at the campus store here for a semester and a half. On average, the Macintosh price premium is around 50% over a comparable PC once you include the monitor. Most of the time, you can get a PC and a Mac with comparable specs at the same price, except one difference - The PC has a 17-inch monitor, and the Apple will need one added ($350-450 depending on how Apple is feeling at the time.)

    Oh, and this is including Apple's 10-15% educational discount vs. the 1-2% on PCs.

    That discount right there says how badly you're getting ripped off - PCs only offer 1-2% because that's pretty close to what their margin is.

    I'm speaking mainly of Gateways here - Dells were routinely 5-6% more expensive (maybe more), and while we carry Compaq, I don't think the store has sold a single one since last summer - all the salespeople would feel too guilty. :)

    An example:
    Gateway E-4200 550
    PIII/550
    128M RAM
    20 gig HD
    17-inch Trinitron monitor
    DVD drive
    Zip drive
    56k non-winmodem
    ATI Rage 128 (Can choose "no video" and put in your own if you don't want it.)
    3Com 10/100 ethernet
    MS Office 2000 Pro (Academic)
    8-bay case
    $2166

    Apple's "Faster" G4
    Same as the Gateway, but:
    450 MHz G4 (Dunno how this compares with the PIII-550 - I'm just going to say "even")
    No monitor
    No MS office
    Case with 3 ext. bays and 2-3 internal bays.
    $2249 - Add the monitor and Office and your system is around $2800-3000.

    Well, I guess I was wrong, the factor is less than 50%, but it's still obscene.

    And let's not forget that Apple intentionally engineered the firmware of G3s not to boot with a G4 processor despite the fact that a G4 was physically/electrically compatible. This is completely contrary to PC mobo manufacturers releasing BIOS updates for the latest processors.

    BTW, I do like the G3/G4 case design - it's just NOT worth the price premium!

    Oh, and one note, as to 3Dfx boards on a Mac - Last time I checked (2 months ago or so), the drivers only supported PCI boards despite the fact that the Macs had AGP video by then.
  • If you're looking for a Linux laptop, get an iBook or Powerbook. iBooks and Powerbooks, unlike desktop systems, ARE competitive in price with PCs.

    iMacs also are feature-for-feature, until you take into account the lack of upgradability.
  • But do I have to buy it bundled with MacOS even if I only want to use linux? Just imagine the cost for building such a cluster... Wasn't motorola trying to sell OSless mobos and got stopped by Apple or something? And what's this about the bios being built kind of to stop people from booting something else? No wonder people building linux clusters go for x86s or alphas... Oh well, this has already been said a zillion times, I guess... Still, nice to see linux running on another plateform even if it's not the most cost effective one.

    ---

  • 1) Ejecting a disk.. all the fun ways:

    Click on the disk press command-e
    Click on the disk press command-y
    Click on disk select put away fomr file menu
    Click on disk slect eject disk fomr the Special Menu
    Drag disk to the trash (for some reasone the most poular)

    Click on disk press command option e (ejects the disk and leave it behind for disk-to disk copy wiht only one drive.. Try it one time.. you'll get to do the disk swapping fun)
    Command shift 1 (same eject and leave behind, note: commane -e does this in systems earlyer than MacOS 8.0....)
    Hold down option and select eject disk.
    poke a pin in the little hole to the left of the disk drive. note: Don't do this unless the disk is stuck.

    Now.. You have all these ways... isn't ONE of them intuative for you :-)

    While the mac can have pre-emptively multitaksed threads. It mostly uses something known as co-operative multitasking. This means the currently executing program gets to decided when it realease the processor (great for games, bad for servers :-) ..).. This exists because the MacOS is a continuous opperating system going back to 1984 when a Mac only had 128k of memory and started up form a 400k floppy. In order to have programs that work on the new system and the old they invented this system of multitaksing. The 1984 bagae is also another reasone why modal dialog boxes block the CPU and freeze the interface. Apple has actually been buggering about wiht this problem ever since it first because apparent in system 6. In sytems greatter than 8.0 there's a set of additional routines for having dialogues that allow you to swithc tasks. Try any modern program that uses the Nativigational service API for opening and saving files and you'll see them. This is very multi-taskable. You can also use MacOS 9.0 for any lenght of time and see other types of dialogues that even allow multithreading. And the MacOS 9 Finder's famous "pasive dialogues"...

    When you see the old style dialogues this is either because the program was made to run under system 7 or sometimes system 6 even or just hasen't taken advantage of the new API because the programmers don't want to since nobody really cares much about this issue outside of people using the MacOS as a low-volume server. This si a bit annoying for powerusers.

    Anyways.. The next version fo the MacOS (version 10 or X whatever) is based on a Unix Micro-kernal based thing so it's completely different. You should read some of the things they're doing to allow programs of today to run on it... They have three APIS sets in there... The classic, Carbon and the re-vamped openstep API based on the ideas of objective C. it should be interresting.
  • Well, PowerPC is the name of the chip so I'm not exactly sure what you're talking about. A chip being open? huh?

    Apple is the company not allowing the BeOS to run on the G4.

    There's other platforms that use PPC.. I'm thinkig baout CHRP and PReP... Of course, these are hardly contenders.

    It's a Pity, because the BeOS's multimedia focus and the G4 are made ofr each other.. The price of politics..
  • Your comments were quite fair:
    • Indeed, there are no cheap, commodity motherboards.
    • This lack is important.

    I'd love to see the OpenPPC Project [openppc.org] do something to provide sources of PPC motherboards that don't cost thousands of dollars.

    Unfortunately, what I'd really want to see is a mobo that costs only a few hundred dollars, and which allows hooking up a couple or four PPC chips. And it looks like there's not going to be any combination like that any time soon.

    Remember: The Intel alternatives may not be "pretty" hardware, but they do make for a compelling lowest common denominator. I can head to Aberdeen [aberdeeninc.com] and locate an SMP Pentium III motherboard costing a couple hundred dollars, toss in a couple CPUs, and have some reasonably powerful hardware for about $1000.

    For PPC to provide a realistic alternative, it needs to either:

    • Provide compelling amounts of computing power that are unavailable on Intel;
    • Provide reliability unavailable on Intel;
    • Provide a price that is not too much more expensive than an Intel-based system of similar power.

    Feel free to s/Intel/Athlon/g or s/Intel/Alpha/g as needed.

    The critical point here is that if the PPC system is outrageously more expensive than an IA-32 system of relatively comparable power, it just won't sell. There are some that are sufficiently bigoted against Intel hardware that they'll pay more for something else, whether PPC, Alpha, or SPARC. I'm not going to pay a $1500 premium to run the same code, recompiled for PPC in order to have a PPC label on the CPU that may not be visible in any meaningful way unless I put a sticker on the case.

    By the way, you may not be quite right about the "only" PPC motherboard being from Motorola; Cogent Computers [cogcomp.com] appears to have one that costs around $1200. Of course, the CPUs to toss into it seem to add another $1200 or more in price, so I could be off here...

    I'm certainly with you in being disinterested in "buying a Mac and ripping its guts out."

  • Good deal? I still can't buy a Mac that can do more than 1024x768 whereas I can buy a Dell PIII 650 with a 15.4" screen that can do 14xx (and looks good).

    Besides, more PC laptops will be integrating 3d support into their laptops soon which means good Q3/Unreal performance even when I'm over at my parents or something (their little 200MMX just can't handle it ;> )

    Apple better get on the ball with their laptops. They seem to consistently be behind with portables now.
  • I can pick up an ALPHA motherboard for less than a PPC one from any one of half a dozen vendors. SMALLER market!

    The real truth is that IBM and Motorola fucked up when they let Apple determine the market for PPC systems.

  • ATX formfactor. That's all, nothing fancy. I'm not interested in buying Macs and striping them.

  • Now, if it was only possible to actually buy one.

  • Vapourware, like the rest. Nothing shipping. I've looked everywhere and the story is the same. Everyone promising 'Real soon now' for the last 6 months.

    I can go out and have an Intel based board in my hands within 30 minutes.

  • I want a PPC based ATX form factor motherboard with the latest CPU at similar prices to an Intel based system. I don't care about Firewire, USB, sound, disk or even a video card.

    I don't want Intel, SPARC or PA-RISC either.

    Jeez, read the posts.
  • by Tsk ( 2863 )
    Some good news show come really soon now from IBM and it's pop architecture. PowerPC might be the firts open platform : see openppc.org for more details.
  • oups missed the link : openppc.org [openppc.org]

  • This is one of the funniest posts I have *ever* read on slashdot.

    For the love of God, moderate this up!

    --Lenny
  • I think that is a valid argument against Macs. The original poster may not have presented the idea in the nicest way, but I agree with his sentiment.

    Have you used the hockey puck? It is uncomfortable. Why ship millions of crappy mice (unless you expect to sell more mice on upgrade.) Yes, it does looks cool.

    And does the MacOS support extra buttons when you do upgrade the mouse? I thought MacOS was a single consistant (dumbed down) interface so you wouldn't be confused by right/middle/left clicking.

    As lease Macs ship a mouse with a button. Maybe the next "innovation" will be the 0 button mouse. Just move the cursor around the screen...

    BTW: -5 troll. Thanks-

    ed
  • Re: #1
    They're pretty affordable for people who buy brand name hardware (which is what it is)

    For a while there were parts floating around so that you could build your own Mac - I did - but when the cloning era ended so did that. And even then the reason wasn't so much the price (which was pretty similar) but that Apple didn't have BTO at that time.

    But it makes no sense to compare a brand name machine with one you make from generic components. None of them are going to ever be as cheap, and they're really two different markets.
  • Well there are a couple of other issues as well, but yeah, good PS drivers is a big one.

    Additionally, besides having exceptional mindshare in the DTP market, Macs just have a lot of software already developed for them that fills small niches but which no one is willing to give up.

    And of course, while this is difficult to describe, Macs have a different 'feel' than IBMs. Even moving the mouse doesn't feel the same. Once you get comfortable on the Mac there's not much incentive to leave it. It's like the comfy chair of computing ;)
  • Regarding Office, why would MS bother giving it away? They already have an OS monopoly, and there are virtually no competitors with Office on the Mac. Don't be suprised if they start charging for it again someday on the IBM side.
  • heh. Mac users either install drivers off of CD (which is probably cheaper than floppy by now) or don't need 'em at all, because it just works the second you plug it in.
  • If Be cared they could be running on current Macs - the truth is they don't care and they're just badmouthing their competitor. The various Linux folks have succeeded, some with Apple's support, others with none, but the results of all of their work is open for Be to study even if it can't be bothered to do so on it's own.

    The simple truth is Be ported to Intel, then got invested in by Intel (sits on Board of Directors) and is now moving away from from the Desktop OS market altogether after a decade of never really succeeding. Their giving away of BeOS-lite is simply a publicity stunt (what, they're going to *sell* lots of copies after they've announced they're dropping the product?) and an attempt to mollify the developers they've burnt.

    Be's inability to move to the moden Macs speaks more about Jean-Louis Gassée and his team then it does about Apple. It also speaks very loudly about Be's enthusiasm for disinformation and blaming everyone else for their lack of success.

    Here's to the little guys and gals who succeeded where Be couldn't go.
  • Well not really too flavors of *BSD on PCI PowerMac computers. Only NetBSD runs on PCI PowerMac computers and some RS/6000 models, OpenBSD requires a RS/6000 or CHRP PowerPC model (due to problems with the PowerMac's OF).
  • Also don't leave out a few groups that are toying with porting to the PowerPC, namely Slackware and Mandrake. The last time I had heard, there were 3(!) teams working on porting Slackware to the PowerPC, including the offical team who owns a iMac and a G3. Linux-mandrake employees John Buswell, one of the orginal/leading developers of Linux for iMac, and runs the site iMacLinux.net. While the company apparently is toying with it internally ("because everybody else is doing it"), they have no offical release planned yet.

    I am currently writing this using Debian/PowerPC that I install one Friday eve, about a month back. While it's far from perfect, it's a quite nice distro, and it's alot easier to install then people make it to sound. It also seems about as stable as YDL, and much more reliable then LinuxPPC. apt-get, and all of the debian tools are really nice.


  • i've been using linuxppc since version 4 came out, and i can't praise it highly enough.

    i can see that the long-term view requires photoshop for linux - and hey, they released it for sgi - but the real strength of the ppc port to an old mac head like me is that when you upgrade your mac (and you do, you do), you get a free webserver in the old one.

    my phone bill is a tiny fraction of what it once was, my productivity is vastly increased and my ability to make things happen online is far greater than it was because i know how to make the box bend to my will.

    i've got the websites of twenty of thirty large NGO's sitting in an old mac clone here, and all for nothing.

    well, for a hundred bucks a year, because i subscribed, but.

    thank you.

    will

    ps. need any help with linuxppc.org?
  • better yet, "endian-little" or "ianend-tlelit"
  • by Rozzin ( 9910 )
    "Apple refuses to allow Be to get the specifications necessary for BeOS to run on the G3/G4 processors."

    That's interesting, considering that Apple doesn't own those processor specs; the PowerPC chip design belongs to IBM--aside from the fact that Macs now contain them, what does Apple have to do with PPC?
  • I haven't tried LinuxPPC yet (although I fully intend to), but I did Linuxify a couple of old Mac 8100s using mkLinux and it was a suprisingly pleasant experience. Once you get over the shock of seeing white & black console on a Macintosh screen, it is actually a speedy setup. X11 seemed to be much speedier than MacOS 7.5 (I'm not flaming MacOS, but you must admit that the newer versions are VERY slow on older hardware) and almost everything seemed to compile with no hangups. So if you ever come across a PowerPC when dumpster diving (or offer to help "haul away" some that are being replaced by Windows machines as I did), I strongly recommend giving Linux a try. Penguins like apples too!
  • Just a quick note to further thank Troy, Dan and the rest of the folks at Terra Soft for their continued interest in supporting non-Apple PPC machines, and to encourage people to check out this project [openppc.org] to find out about the future of non-Apple PPC boxen.

    I can hear the yowls from here... "What non-Apple PPC machines?" Right now, there aren't many other than IBM's server line, although you can buy a Sandpoint reference board from Motorola and miscellaneous boards from smaller vendors. (See http://www.openppc.org/vendors.html [openppc.org] for a full list.)

    But back in August, IBM announced that it would be releasing an open, reference motherboard design for PPC ("POP", or PowerPC Open Platform). The schematics came out quickly, but the whole project's been bogged down in testing because of a faulty northbridge. In the meantime, however, a small community project has sprung up to track POP, at http://www.openppc.org [openppc.org]. Interested parties are encouraged to sign up for the mailing list [phys.sfu.ca] and plumb the Web site [openppc.org] for tasty nuggets of goodness.

    --Tom Geller
    Co-founder, The OpenPPC Project

  • Running the Distributed.net rc5 client on a G4 400 MHz (MacOS 9) with the altivec core was cranking out 3.5Mkey/s. Running the 'same' client on a Pentium III 500 MHz (Win NT 4) is storming along 1.4MKey/s. Which would imply the G4 is about 2.5 times as fast, while running 20% slower clock.
    Now all I need is a dual G4 system with linux pre-installed, a linuc rc5 client that uses altivec, and getting rid of those noisey fans would be nice...
  • why do we keep making cleaner architectures and faster chips if we're going to sit around spending all our time to trying to make compilers spit out hand-tuned machine code?

    The original point of RISC was moving complexity out of the hardware and into the compiler. Of course the hardware is still going to be doing crazy branch prediction and whatnot, but the compiler should still do its best to keep those pipelines filled.

    And hey, if you can make a program run faster without having to dip your mobo in mineral oil, why not?

  • The difference between little-endian & big endian is that on a little endian machine:
    int x = 1;
    char c = *((char*)&x);
    Will assign 1 to c. On a big endian machine c would be 0. By the way, the line just assigns the lowest byte of the word x to c.

    This can cause major headaches if you are doing bit-twiddling or mucking around with funny casts.

  • Because for the vast majority of Linux users runs it on a x86 PC, so submiting patch for a machine that you don't use is quite difficult...

    If you are talking about "endianness patch" then I agree: IMHO code which depends on the endianness are pure evil.

    On a side note, how good is GCC on the PowerPC ?
    On one hand, the numerous registers, the "normal" floating point unit of the Power architecture should help the optimiser, on the other hand there has been much more effort to tune the GCC compiler to the x86 CPUs than to the Power architecture.

    Just wondering...
  • You left out Solaris for PPC, NetWare for PPC, OpenStep for PPC, and probably a few others.

    The big predictions in the early '90s was that Intel's architecture was too complex, and in the long run Intel wouldn't be able to scale. People produced nice powerpoint graphs showing Intel leveling off to a flat line while RISC chips continued their linear speed growth.

    So, on the surface, PPC seemed like a good idea as a replacement for x86 PCs on low-end commodity hardware. The problem was all of those damn x86 applications. And that, so far, Intel and AMD have been scaling just fine, for some periods even beating Moore's Law. So while PowerPC has been a great chip, there's never been the enormously huge returns that would encourage people to switch architectures.

    OS/2 wasn't exactly super popular even on Intel hardware, so no big shock the PPC version was killed. WinNT hasn't displaced Windows 9x yet even after 7 years of trying, and even on Alpha where the thing was fast, it wasn't exactly flying off the shelves. And Apple wasn't letting anyone port their OSes to their hardware anyway.

    If LinuxPPC, Inc. was a commercial enterprise in the sense of Microsoft or IBM, their product would have been killed by the beancounters a long time ago. Fortunately, the LinuxPPC project is driven by volunteer hackers who just want to run Linux on their Mac hardware. It doesn't need to be 'viable' commercially.
    --
  • Power Computing did not make 'open' PREP or CHRP machines. Their machines used Apple designed and Apple licenced boards with the Apple ROM. They were as closed as Apple boxes, because they essentially were Apple boxes. As for destroying jobs, I doubt it -- Apple's market share is now greater with one manufacturer than when there was dozens of clones.

    Other companies like Umax made both Apple and PREP machines, as well as bare PREP boards.
    --
  • One thing to realize about the PowerPC platform is that it was intended to be a cheap, commodity platform just like Intel. It was not intended to be a high-end workstation platform like Sparc or Alpha. So, you are right -- it doesn't have an overwhelming price/performance ratio -- it was not intended to.

    Now the main problem was that PowerPC as a commodity platform is that nobody bought it. If people had, there would be price-competitive PowerPC-based solutions available. Several manufacturers had PowerPC based systems and bare boards on the market a few years ago -- the platform just plain failed to make any money for them.

    You guys seem to be putting the cart before the horse. Intel solutions are not popular because there are cheap loose parts available that you can roll your own with. Intel solutions have cheap parts because Intel solutions are popular.

    If you think that cheap, bare PPC boards are going to ressurect the platform, think again. For every x86 motherboard sold to an individual, Intel sells several hundred thousand chipsets to Dell and Compaq and Gateway. The fact that you can get your hands on a $100 motherboard is a pure economy of scale issue. The guys at my Korner Klone store have no interest in PPC-based solutions. Their bread and butter is Windows 98 and NT and Novell, and they buy a hellava lot more motherboards than you and I put together.

    Now, since this article is about Linux-on-PPC, why is this a flamewar? You have a source-compatible OS, so go and run it on the best damn processor you can find. If that's a homebrew x86 machine that's great; if it's an old Macintosh that's also great. Just don't expect something (a homebrew PPC machine) to come into existance just because you think it might be cool.
    --
  • It's substandard as a gaming card, sure. But most Mac G4's get put into use as 2D graphic workstations with big high frequency monitors, and ATI puts out some of the best 2D image quality this side of Matrox.

    I notice that Dell ships either the same Rage128 or a TNT2. If I was looking at photoshop and quark all day, I'd take the ATI. Nvidia's 2D is not-so-great.

    Apple should be shipping the home-use iMacs with Voodoo cards, however, and they also should be an option on the G4.
    --
  • This was a nice little article - it's great to see some high-profile recognition for Linux on one of its non-x86 architechtures. I've been using LinuxPPC since the 3.0 release in February 98. These earlier libc5-based versions of LinuxPPC were tough to deal with - amost nothing ever compiled out of the box, XFree86 didn't work, compiling kernels was an adventure and performance was slow. (At least this was my experience). I use Linux on an Intel machine at work and have the "1999Q3" release of LinuxPPC on my home computer, a PowerMacintosh 7600. The two compare favorably. Now glibc based, compiling important Linux apps is almost never a problem with LinuxPPC. For testing and development, it runs Apache 1.3.12 with mod_ssl, PHP4, MySQL 3.23.14, kernel 2.2.14, etc, etc, etc. Suffice it to say that it *can* be used for "serious computing" and is fully up-to-date with the latest kernel and apps. I also find that Linux makes efficient use of its 132 MHz 604 processor, and is plenty fast. Since the change to XFree86, X runs *much* faster on this computer. The only major problem, as far as I'm concerned, is that GNOME is painfully slow. Since I don't use GNOME apps that much, though, this is not a problem for me. If you tried LinuxPPC a while ago and were underwhelemed, give it another crack now. It's improved by leaps and bounds and you probably won't be disappointed unless you simply don't like Linux on any architechture. Mac users who consider themselves "power users" should definitely try it. I still use MacOS on occaison, and in fact still like it quite a bit - I grew up with it. But so many Mac users are stuck in the rut of thinking there's only one way to do things and that theirs is the best. I know because I used to be one of those people! LinuxPPC shows that "there's more than one way to do it" and can help you gain a better understanding of computers, regardless of what OS you choose to use day in and day out.
  • For starters, don't write makefiles that assume things like -D486 for compile flags. That is incredibly annoying for scripting builds. Or if you absolutely feel the need for such flags, make sure that you can detect architecture and have the option for an appropriate flag (like -DEV6 on certain Alphas.

    Also, if you're going to deal with bitwise operations, remember that some machines are big endian and some are little endian. You'll probably need to write different operations for each type.

    Finally, Alphas, UltraSPARCs, and some other machines are 64-bit systems. Intel is 32-bit. That will get you into trouble at times.

  • I run a website at http://www.macdiscussion.com and did an interview a while back with Jason Haas from LinuxPPC Inc. good reading if you are interested in finding out more about linuxPPC and linux on the ppc in general. Watch for an interview with SuSE ppc developers coming next week discussing their ppc port and linux on the ppc... This is not a 'dead' chip...slackware is now also porting to ppc, and there are at least 2 flavors of BSD running on ppc as well...
  • According to a test in the German CT a year ago a G3bw at 300 Mhz had a sustained transfer speed of 50 Mbyte/sec with Gigabit ethernet.
    A very expensive Intel machine with 64 bits PCI (like the Apple g3bw) could only transfer 25 Mbyte/sec sustained with OS/2.
    With NT and Linux the transferspeed dropped to 22 Mbyte/sec.
    Technically the Mac has a lot better IO-system.
    O yeah, the Mac was running MacOS 8.6.
    And with 100baseT and average Mac (G3 233) can outperform every Intel based computer.
    Due to the fact that every intel clone has to be backwards compatible with the first IMB XT the I/O performance really lags compared with Apple Mac's, Sun sparc's and Digital Alpha's.
    With the introduction of MacOS 8.6 there was a problem with different ethernet switches which couldn't keep up with the Mac.
    It is possible to get 100Mbit/sec sustained without a problem.
    Every Intel PC I know has a limit of only 86 Mbit/sec. with 100baseT.
    This has something to do with interrupts/sec.
  • The LinuxPPC and Yellow Dog guys make money the same way Red Hat and Debian do -- support. Debian do not make money, even if some developers may support something for money -- Debian itself is just a voluntary, non-profit organization.
  • I haven't used the one button mouse for years. My Turbo mouse is ugly, but beautiful. When I get a powerbook, I want the little two button USB mouse by MacAlly. It fits right in my shirt pocket. My G3 Mac will soon be a SuSE machine when my G4 arrives.
  • by Anonymous Coward
    The thing that most everyone forgets about that sets the Mac apart form Windows with regards to Graphic Arts is postscript. I run a prepress department with all 3 major OS represented. We have Macs, Windows98, NT4.o and red hat Linux. We can't even begin to think of using Red Hat as a DTP platform, but what a great file, ftp, intranet server! Window is more stable than Mac and we use it for fileserving, and Ripping (turning postscript into dots). When harleqin has more available Linux ports we will probably go with those over NT versions. What nobody does as well as the Mac is write good clean postscript. That is an OS thing, not a hardware thing. Windows postscript is not as reliable as Mac postscript and in the long run we waste far more time on unreliable postscript that we do on crashing OS's. That is why Macs are still used in Graphic Arts more than PC's. PCL is not capable (in my experience) of doing the complex graphics work done daily in graphics art houses. (Maybe it is capable and is just poorly implemented, I don't know. I only know that in my daily use it doesn't cut it) . Postcript, for these people, is the only way to go and that means Macs or Sun (they do postcript well too). That is what you will find in real graphics houses writing the postscript code. Once the code is written, anything goes. Mac, PC, Unix and more and more frequently ->Linux. (Yay)
  • This thread is about LinuxPPC.

    i.e. Be has access to all the information they need to to get BeOS to run on recent Macs.

    While they can't openly copy the code, nothing prevents them from reading it to find out how the hardware works.

    I'm inclined to agree with those who mention that it's an issue of marketing - Be just needed an excuse for dropping PPC hardware support.
  • Intuative simply means: works as expected; or is consistant wiht other things you've already experienced. In thia way it's possible for a computer to be intuative. All it has to be is internally consistant and use metephores from the real world or other situations the user might have already come into contact with. These two ideas work together to provide intuative behavior.

    1) Self consistancy: The MacOS has traditionally worked to build a consistant enviroment.. ie: all programs have (or should have if they follow MacOS interface guildlines) a FILE mane and an EDIT menue with a QUIT and CUT COPY PASTE menue items irrespective of weather they even use them or not. There's standard keyboard shortcuts of Comamnd C for Copyu, Command W for close window and the best: comand Q for quit. This means you can start up a completely new application you've never seen and type in these keyuboard shortcuts (among many others) and know what the program will do. This goes to the very deffinition of intuative. All aspects of the MacOS try to do this. In fact things like the Quicktime movie player and "Sherlock II" have been reviled by most Mac Power users since they do not use the standard interface or interface behaviors. Also you'll note that when you drag something somewhere it's either a move, or if a move is not possible, a copy.. The MacOS doesn't have the (as far as I can tell) completely random behavior windows exibits whenever I try to move or copy something.. (A shortcut? Did I tell you to make a shortcut.. no, I said move you stupid thing.. oh so now you're going to move.. and what about now.. a copy.. great..)

    2) Use realworld metaphores: I'm thinking specifically of things like the desktop, although this can also be found in the icons: use a pencil to draw.. Use something that looks like a notbook to store notes (notepad). There's a folder here that has a mini computer on it.. Oh that must be some sort of computer code... (system folder)... What about this document? (It's a word doc) it looks different than this document (appleworks doc),.. anwyays...

    Making a computer easy to use typically is a question of making it intuative.. That was the original point of a GUI.. although now it's degenerated into more of a "scares user less than a command line" approach where the user is some sort of baby.. oh, the user drags to the desktop.. he really wants to make an alias.. we'll do that for him. Who's supposed to be in control here? This is the kind of thing that windows users have toruble with when comming to the mac.. the mac doesn't try to baby the user. it does what you want and what the user should expect should happen when they do that. anyways...

    As for things like protected memory.. this si only usefull to programmers.. Most users I know of never use more than one app at a time so don't care if only one app goes down... At leats the MacOS tells them they should restart their computer now.

    Pre-emptive multitasking. It would be nice but in typical usuase I never run into this problem... The MacOS multitaks just fine 90% of the time. To me, having a nice intuative interface is more usefull then having pre-emtive multitasking.. It's more powerfull. I can use any generic program in existance with hardly any learning curve. If you really want to see how having a good interface is powerfull check out the old Claris programs... Very strraight forward apps that any newby can use with some veyr powerfull abilities.. In the old claris works you can dump a Spreadsheet into any kind of other document.. for that matter a draw or paint or word prcessing doc too.. Sort of like OLE, but it runs on an 8Mhz mac SE, takes up 2 megs of ram and 2 megs of disk space... The best part is, it doens't interfear with new user's. You have to click on this button down near the zoom in- zoom out buttons... Anyways.. now that Apple has screwed all this up in AppleWorks 6... I'm sticking with 5.0...

    As for the File System and other bottlenecks.. yes, these are a pain. I expect these are problems programmers are up against when they try to make a big-full featured GOOD app on the Mac. THis is the real reasone Apple is going to MacOS X.. It's better to program on. I mean, having to restart your computer evertime a program seg-faults is insane! AGGGH! And that's if it dies right away.. sometimes it only makes your system really unstable so it crashes in code that was corupted by your code.. anyways... MacOS X is for developers. I'm learning Objective C right now because I figure MacOS X will be the ultimate developement platform... :-) bye
  • The ONLY shipping standalone PPC motherboard you can buy is from Motorola and it costs $2500. This is FACT. Don't believe me? Have fun searching.

    I have absolutely no interest in buying a Mac and ripping it's guts out.

    I just want a PPC based motherboard without all that Apple bullshit.

  • There are too many naidne-elttil x86-only folks out there...

    Surely that ought to be ilttel neidna? :-)

  • I know that the PPC is big-endian and x86 is little endian (unless I got that backwards AGAIN). How much of a pain in the ass is this as far as porting code goes? No too bad, as it happens, at least for userland code. I have no experience of kernel level code. I've been running Linux on Sparc hardware (big endian) for nearly 5 years now, and have encountered surprisingly few problems. You do get the occasional jokers that ship source code with inline x86 assembly. Most are fairly reasonable about it, and are prepared to work with you to get a C equivalent written. All it needs is an autoconf check for architecture to determine whether to use the assembly or C version.
  • Be's position on BeOS on the G3/G4 has always been that the obstacles are political, not technical. They're not going to put resources into being on a platform whose owner doesn't want them there. Apple has made it clear that, despite the fact that they make most of their money from hardware, they see BeOS as competition. (For proof of this, look at the shareware BeOS program "Postmaster," whose future is in limbo now that its author works for Apple--they explicitly told him that they do not want him selling a product on a competing platform.) Could Be make BeOS work on the G3/G4s? Sure. Despite all the ire directed against them, it's important to remember that they've never claimed otherwise. They blame Apple for not supporting their efforts. Can you blame Apple, instead, then? Maybe; they're arguably being bastards on this point. Neither Steve Jobs nor Gassee are exactly known for their shy and retiring manners, though, and the fact is that BeOS wasn't ported to the Power Mac platform from the BeBox... it was ported to the Power Computing platform. Yes. Power Computing gave the support to Be initially, not Apple. The Amelio-era Apple just wasn't hostile to Be. Ultimately can you blame the conflicting egos of Gassee and Jobs? Probably. Life goes on. Personally, I'm kind of interested in LinuxPPC. I'm curious about getting MacOS X on my iBook, but I think there's more than a slight chance that OS X will prove to be a resource-sucking, underperforming swamp hog (albeit one with a lickable snout).
  • An intuitive interface, to those who grok the Mac way, is a powerful interface.

    Of course, there are some of us who prefer a minimalist interface which we can expand as needed with whatever other functionality we require.

    - Jeff A. Campbell
    - VelociNews (http://www.velocinews.com [velocinews.com])
  • I know that the PPC is big-endian and x86 is little endian (unless I got that backwards AGAIN). How much of a pain in the ass is this as far as porting code goes?

    Well, technically, the PPCs have the ability to run either big endian or little endian code. Normally, they are big endian. IIRC, there's an asm mnemonic that allows you to switch the endianness when in supervisor mode.

    As for porting code that relies on a given processors endianness, it isn't that hard in a language like C, which the majority of software is written in. In fact, you need to do this sometimes in software anyway, since some file formats for sounds and graphics (among other things) specify the endianness and other byte order issues for storing larger than 8-bit values. (The whole world is not like UNIX and they don't all treat files as streams of bytes.)

    To convert the data, you write a cracking macro to swap the bytes. While it is trivial, I leave it as an exercise for the reader. A couple of other macros come in handy when dealing with binary data larger than 8 bits, such as macros to get the high order byte and low byte from a 16-bit value, and others for doing various manipulations on 32-bit values. It should be noted, too, that some architectures do rather bizarre things with bit orders, other than just the byte order of larger values. These, however, a not a concern when porting x86 code to the PPC.

  • Some good news show come really soon now from IBM and it's pop architecture. PowerPC might be the firts open platform : see openppc.org for more details.

    While open hardware specs are a Good Thing (tm), openppc's mobo design is NOT. Please turn to page 24 on their PDF file of the mobo design.

    Can we declare ISA dead once and for all? I mean, I know this is IBM and Token Ring really only runs well on ISA, but can we have a MODERN motherboard designed in this decade please?

    How about USB for keyboard and mouse rather than the dated, lame, ISA?

    Two IDE controllers? How retro.

    This is nothing more than a stock x86 design (and all its legacy cruft) with a spot for a PowerPC chip.

    I mean, the thing even has a game port.

    _Deirdre

  • If BeOS actually cared about PPC Mac users, they'd hire people to do the same hard work that they LinuxPPC team is doing.

    The verbal abuse in this statement aside, there are a lot of Mac people at Be. The reality is that Be supports its OS and it won't support something it doesn't have specs to. That seems pretty cut and dried, not to mention a reasonable position.

    Besides, why are you slamming the efforts of a group that has done something you consider far too difficult for Be, Inc.?

    Too difficult? ::chortle:: Not. There's a big difference between being able to do something and being able to *support* something.

    What's so wrong with giving Mac and other PPC users a chance at Linux without having to give up their current machine or get an extra one?

    This seems something to slam Apple for, not Be. I have run MacOS (various flavors), LinuxPPC, Yellow Dog Linux, MkLinux, MacOS X Server and BeOS on various Macs over the last few years.

    _Deirdre (who does not speak for Be or Apple)

  • For the record, Token-Ring runs just fine on PCI, thank you very much.

    For a long time, the drivers for PCI Token Ring were NOT there, especially for Linux. In fact, when I was looking for Token Ring drivers (two years ago), the PCI cards were notably Not Supported. Furthermore, they weren't even well supported on other OSes at the time. I finally used a Linux box with an ISA Token Ring card to use as a router for my laptop (running Ethernet).

    Let's not forgot that USB was designed to replace your serial ports, not the ISA bus.

    Keyboards and mice shouldn't be hung off of ISA. USB is a more-than-acceptable way of doing it and offers advantages of doing MORE than just mouse, keyboard and game port (!).

    USB networking, for example, has roughly 60% the throughput of a even a cheesy $5 ISA card.

    This isn't even relevant to the point as networking is done on this board through PCI (which is the most appropriate solution).

    _Deirdre

  • Well, OK, if you are running a niche application that involves highly parallel data, then yes, a chip with a good vector unit is going to blow away one with a mediocre or non-existent vector unit. On the other hand, if you were to run an application that didn't make good use of Altivec, then the G4 is going to be about the same speed as an similarly clocked pentium.

    This is why benchmarking is so contentious: you can dream up a benchmark to prove pretty much anything. But the fact is that most every-day applications are not going to make as much use of altivec as RC5 or seti@home. So these 3-6 times as fast ratings are simply not credible for everyday use.

    Don't get me wrong: the G4 is a kick-ass chip, and if Apple had more competition, they would almost certainly be a better price/performance deal than x86 offerings. In fact, when I finish being a starving college students, I plan to buy a G4 (or whatever comes after it) I couldn't stand running Windoze or Linux as my only desktop OS choices. But it does the Mac community's credibility no good to treat Steve Job's inflated benchmarks as revealed truth. Yes, the G4 probably beats the competition in photoshop tests. But for those of us who don't do graphics for a living, that's not a reasonable benchmark. Like every company, Apple exaggerates their product's benefits.

    Now that IBM is making G4's, though, I'm looking forward to seeing quad GHz G4's running OS X hit the market in early 2001. *drool*
  • The PowerPC architecture will always lag behind overall, it's just that its very close at this point in time.

    The current high-end G4 is 500 MHz. That compares with the 1 GHz chips Intel and AMD are claiming. I'm skeptical about whether those chips are going to ship in volume any time soon, though. Apple claims that the G4 is over twice as fast as a similarly clocked pentium. This is probably a bit of an exaggeration, but I think it probably is true that a G4/500 is equivalent to a P-III/750 or so.

    So in terms of raw speed, the high-end x86 are probably ahead. But the G4 has a *lot* of room to grow. If the folks at Moto weren't so incompetent, we'd have 7-800 MHz G4's by now, which would blow everything x86 out of the water. And IBM is just now ramping up G4 production, so expect to see some serious increases in clock rates in the coming months.

    So it's ridiculous to say that PPC's will "always lag behind." In fact, the exact opposite is true. They are currently in something of a slump, and with a little luck they will catch up and surpass x86. The PPC is simply a newer, cleaner architecture with more room to grow, and so they shouldn't have any trouble keeping up with the kludge that is x86.
  • If it was your $400 Million, which would you have purchased -- BeOS (still in Beta), or OpenStep (at version 4.2)?

    I mean, I like BeOS and all, but NeXT was just a far better deal even ignoring the personnel factor (especially because it included a bunch of dev tools and WebObjects).
    --

  • WTF? The same damn troll post that's on every story mentioning a mac, and it's Insightful? Come on, I hope no moderator is stupid enough to believe that macs can only handle a 1 button mouse (mine has 4 and a wheel...)

  • Be careful putting your mac partition in read-write mode... Every time I do that and boot mac-on-linux it wipes out my desktop folder...
  • Nah.

    2 and 3 button ADB mice have been around for years and often work right 'out of the box' with LinuxPPC.

    2 and 3 button USB support seems good too - its what I'm using. (USB is backported to the 2.2 kernels on PPC).

    If you do only have a 1 button mouse, you can set up a couple of keys to emulate the other buttons...
    Lord Pixel - The cat who walks through walls
  • I believe (somebody please correct me if I'm wrong) that the PowerPC people are reverse engineering what they need, without any help from Apple (the chip isn't the problem, it's the motherboard - Apple won't release those specs).

    The BeOS team has decided not to reverse engineer everything and then have Apple change the specs and break everything again.

    In short, it could be done, but it's Apple, not Be, that is preventing it.

    LL
  • What are you talking about? G4s start at $1599.

    400MHz G4
    1MB L2
    64MB SDRAM
    10GB Ultra ATA
    DVD-ROM/DVD-Video
    RAGE 128 Pro
    10/100BASE-T
    56K internal modem

    Ram & HD upgrades you would want to do yourself, cause Apple charges a bit much. You can probably find lower prices if you shop around, that price is straight from the apple store. A lot of other places will have the same prices but throw in extra ram or something.
  • The Macintosh follows the desktop metaphor. What do you do when you're finish with something on a desktop?

    1) Stick it in a drawer or other container. That's already reserved for filing files and data.
    2) Discard it from the workspace. In other words, the trash can.

    Why no eject button? It's very simple. You do not want users taking out disks in the middle of an I/O operation or before another program is finished with it like you could in DOS. It was an attempt at protecting uneducated users from harming themselves and the system, especially back when the OS booted off a floppy.

    To be honest, I agree that the trash can is better reserved for erasing the disk, but there is no other good item in a desktop metaphor for removing something from your workspace completely. Most other items I can think imply the ability to reach in and get the info back.

    As to the multi-tasking question, I believe Pinky's response handles that question admirably. Another point about the old system dialogs is that the old Macintosh used to be a single-tasking system like DOS. Only one program could run at a time. When they went to multiple programs, a modal dialog is supposed to represent something you DON'T want two programs potentially messing with at the same time -- something important enough to interrupt the user for. However, there wasn't a clean way at the time to make application-specific dialogs for those messages that should only stop one program. As Pinky points out, though, this has been fixed.

    I would like to know in what way either of those examples crosses off "powerful" from the list. Just because awk, sed, and grep on UNIX are horridly counter-intuitive at times doesn't make them less powerful. Just because something might be a little inconvenient doesn't mean it isn't powerful. The UI is powerful because of the wide variety of things you can do with it with little effort on your part. As a whole, the GUI is much cleaner and intuitive than it's competitors.

  • Perhaps you underestimate the niche market. Just look at how many companies are alive making Amiga products still. OS/2 dropped support for PPC because they were already starting to lose profitability in the x86 market. Maintaining two platforms was more than they were willing to fund. PPC NT was dropped for the same reason that Alpha NT was recently dropped -- Microsoft pushed the burden of the development on the hardware manufacturers and charged them huge licenses to continue developing it. In the first case, it was IBM who refused to shoulder the burden after Microsoft put the squeeze on them as a competitor, and in the latter it was Compaq.

    None of these say anything against the viability of the PPC platform, which has incredible applications in the embedded (Motorola) and high-end server (IBM) markets. Unfortunately, Apple's stuck in the middle with the desktop market that neither chip supplier really cares intently about focusing on currently. The LinuxPPC and Yellow Dog guys make money the same way Red Hat and Debian do -- support.
  • Oh, I'm not saying that. There were things for which the Mac was the unquestioned superior platform, but they're growing fewer and fewer as products on Windows come out that whittle the advantage down to being negligible. There are things for which it was never the best for, many of which are strong suits for Linux and other platforms.

    However, if you slam on the interface as being powerless, complex, difficult, and unclean, then you haven't used it in an unbiased fashion.
  • The reality is that Be supports its OS and it won't support something it doesn't have specs to. That seems pretty cut and dried, not to mention a reasonable position.

    You make an excellent point -- one which I had forgotten to consider. However, the primary revenue stream of Yellow Dog Linux and LinuxPPC, Inc. is from supporting these same kind of machines running Linux. My point is that Be, especially in their FAQ, used to beat their chest repeatedly that it is all Apple's fault they are not supporting Apple's machines. However, other businesses are basing the majority of their revenue stream on doing the very thing Be claims Apple's lack of spoon-feeding prevents them from doing. The fact remains that Be is eventually pulling out of the PPC market and is happy to try to turn as many Mac/Be users against Apple to keep as many as possible. They are going where the money is currently and making the decision about where they think they can cut the biggest niche.

    This seems something to slam Apple for, not Be.

    Yeah, in retrospect that seems to be what the original poster was griping about, but I was making a general jab at all the people who immediately got on this thread to flame against supporting the platform, probably mainly because it's primarily the Mac platform, instead of respecting the effort and skills behind the porting work.
  • Do you honestly? I mean, what honest competitive advantage does the system have for you if you're comfortable with Linux and don't want a Mac? The board in question isn't outrageously better than competing boards for embedded appication (in fact, Motorola sells better solutions for that), and the processor is a slower, older model, not as good as most modern desktop systems.

    Besides, a good $1600 G4 comes with a decent video card, a modest hard drive, on-board support for Firewire, USB, UltraATA/66, Ethernet, and sound. The Motorola board has no video, no sound, and no hard drive with it. It's a decent buy, and you can always add stuff to it. Plus, the hardware that comes with it is far more supported on PPC Linux than whatever random peripherals you decide to stick on the Motorola board.

    I don't think you honestly want anything to do with a PowerPC motherboard. You sound like you're just taking the opportunity to bitch about Apple's prices and slam a worthy development effort that seems not to be targetted at you. Well, live with it. There will be users of PPC platforms for years to come, and there will be enough of them interested in Linux to support it. If all you care about it money, go buy an x86 board. Try getting a SPARC or PA-RISC board for less than $2500.
  • by hey! ( 33014 )
    Airport is just WaveLan cards repackaged, no?

    If so, in my naive opinion it seems like you should be able to use the wavelan drivers in the PCMCIA package. Get the latest first, though.
  • FYI for other readers, OpenSSH1.1.2 compiles easy and works fine.
  • As the title said, I was posting my experiences. When I installed YDL a few months ago, yaboot was not recommended; I see that has changed. Hooray for OSS; if we had to rely on Apple or MS to fix something like this we'd wait for years.

    One thing about Macs is that they're nice and quiet; I'd be interested though in any kind of data you have to back up the impression that the G4 is fast running Linux.
  • I installed OSX on this box first.

    The install is very clean and easy. If you want to get a Unix running on a Mac, this is definitely the path of least resistance.

    I took OSX off in favor of Linux because I didn't need to run Mac binaries on the thing and the OSX GUI was not really Mac-ish. It had a bit of a Frankenstein monster feel to it, although I suppose you'd get used to it. The docs werent' that great either - not enough to be better than having access to source code and a gazillion Linux hackers. All in all, I had nothing wrong with it, but it didn't offer me any compelling reason to support yet another operating system (other than hacking value, but my hacking time is totally overcommitted). However it is very intriguing and I definitely will install new OSX udpates as they come down the pike.
  • We've had considerable success with YDL on iBooks where I work. It runs like a rocket (and yes, multiple-button USB mice are supported); we're just waiting patiently for the AirPort support. Coupled with a replacement 12GB hard drive and extra RAM, you end up with a stylin' machine that can run three OSs (MacOS with VirtualPC on one partition, Linux on the other). For me and my co-workers, it's turned out to be a decent development machine. For those interestd in putting YDL on their iBooks, we'll be posting an iBook/Linux HOWTO on our site shortly.
  • For NuBus PowerMacs MkLinux is your only option as far as I know. I use it on my Nubus Powermac. MkLinux doen't get much press but its great for those with older machines. Newer machines won't be left out, but due to the smaller community, it takes longer for support for newer machines to be built in. A new version is due to be out RSN (Real Soon Now) which will provide a much needed update to the libraries. I can't wait for it. There aren't many binaries available, but just as with LinuxPPC, most apps will compile just fine. Typing ./configure, make, make install isn't much more difficult than installing an rpm in my opinion. Don't overlook MkLinux. Check out MkLinux.org for more information.
  • What about BeOS [be.com] on G3/G4 systems? I guess PowerPC is not as open to everyone as Linux is...

    --
  • While I certainly think apple should add another mouse button because it does give users more options, I think that there is something that people coming from windows/linux backgrounds are not understanding about Mac software design. Keyboard shortcuts on the Mac receive far more emphasis in the design of softare than in Windows/Gnome (contrary to popular belief). Mac programmers, arguably the world's most experienced at programming interfaces, assign easily remembered keyboard shortcuts to the most commonly repeated functions. The logic being, the tenth time you do x procedure, you'll look at the keyboard shortcut on the menu and start using that. "But many users rarely use the keyboard for stuff" you might say. My response: it's not surprising. Windows' (and GNOME and KDE's, which both aspire to emulate Windoze down to all it's UI mistakes) keyboard shortcuts are crippled by two things. The first is that Windows cripples any element of ease of use that keyboard shortcuts have by using all those mnemonics (the underline things that you use with the alt key control menus with). These are very good at distracting the user's attention away from the actual keyboard shortcuts themselves. With all those underline thingies already in the menus, the average user is going to feel like it is too much trouble to remember the actual shortcuts. This was originally done because in ye olde Windows days, some people didn't have mice. So the original goal of Windows' keyboard UI was to replace the mouse, whereas from day one, the goal of the mac keyboard UI was to complement it. And that's exactly the function that a second mouse button performs: it complements the basic mouse. Since macs have no mnemonics, there is less clutter around the actual keyboard shortcuts, which means the keyboard shortcuts stand out like sore thumbs, which means the user's attention is more quickly drawn to them, which means the user is more likely to use them. So by using keyboard shortcuts in menus uncluttered with mnemonics, we mac people have been able to get along pretty well without a second mouse button. The second thing that both the linux and windows development community have done to cripple the ease and power of keyboard shortcuts is using the control key for keyboard shortcuts. The modifier keys next to the space bar (the alt key on windows, the command key on macs) have the widest key coverage of any other modifier key. It is ideal for fast, easy keyboard shortcuts because of its good central location on the keyboard and the fact that it puts the two most dextrous fingers of the human hand, the thumb and index fingers, in a very natural position If you don't believe me, put your thumb on the alt/command key, hold your index finger 2-3 inches from where your thumb is resting, and move your index finger in an arc motion that pivots around the thumb on the alt key. You'll notice in that arc that all the keys on that side of the keyboard are covered. If you tried the same thing only with the thumb resting on the control key and still using your index finger, after the first few letters your hand would immediately start bending in weird contortions. I know some people who use their pinky on the control key instead of their thumb, but that is still less natural than the thumb+index combination. PC keyboard design also helps discourage use of keyboard shortcuts. If you take a look at mac keyboard, the command keys will often (but not always)be larger and more centralized than the alt key on a PC keyboard.
    To sum everything I've said, macs have other ways (often more elegant) of performing the same tasks a second mouse button does. Understand this before you bash macs for having a one-button mouse.

    *Note this post does not mean to a flame directed towards Linux. The wonderful open source nature of Linux makes it very easy to rip out the messy mnemonics and control key shortcut sequences in GNOME applications and replace these beasts with uncluttered menus that use the alt keys and appropriately chosen letters. (I've done with with the gnome file manager (gmc) and Dia, among others. I can't wait to do it with Nautilus. I love you, Gtk ItemFactory).

  • while we're on the subject, some of you may recall that linuxppc developer jason haas was nearly killed in an auto accident with a drunk driver a few weeks ago. his wife cassie has been writing in w/ updates on his progress here. it looks like he's alot better, but it'll be some time before he can work on it more :(
  • Several manufacturers had PowerPC based systems and bare boards on the market a few years ago -- the platform just plain failed to make any money for them.

    Well I remember them. They were pretty much the Mac clones. I bought one because I was into the BeOS and I didn't want to deal with the Apple BIOS. What happened to them? I can tell you. Power Computing in Round Rock Texas was making the Mac clones and they were doing well. People were buying. They had record profits. The same for some of the other ones. When Power Computing was getting bigger Apple saw that the clone makers were taking a piece of their market. Rather then competing and letting the platform mature, Apple then stopped licensing the MacOS to the clone makers. This resulted in them losing almost all sales because machines preloaded with the MacOS was 99% of their business. This is the reason why I hate Apple. I know that I am biased but they destroyed a platform and thousands of jobs because they didn't want to compete on a level playing field.
    Molog

    So Linus, what are we doing tonight?

  • Wrong. I can put together by going to pricewatch an Athlon system with all the RAM, case and everything, including monitor for LESS then $1000. $1000 $1700 with no monitor if buying a G4. If you want me to I can post the links directly to the parts I can get and I'm not talking about crappy parts either. This price I am quoting includes 128 MB RAM, a Voodoo 3 AGP, and other standard components. I'm not ripping PowerPC. My father designed the L2 and L3 cache, so I am partial to it. Just was stating an economic fact.
    Molog

    So Linus, what are we doing tonight?

  • o I have to buy it bundled with MacOS even if I only want to use linux? Just imagine the cost for building such a cluster..

    Currently the only way to get a Mac is bundled with MacOS, but eventually, IBM will be shipping their PowerPC open platform boards, which you can buy independently of any OS.

    And what's this about the bios being built kind of to stop people from booting something else?

    PCI PowerMacs use Open Firmware (think of it as a BIOS on steroids.) The problem is not that Apple built it to be incompatable (it was developed by Sun) it's that Apple hacked it up a little bit to make it especially compatable with MacOS, which breaks a lot of things for booting other OSes. You can boot Linux directly from MacOS, though, and LinuxPPC 2000 has a nifty way to direct boot at a slightly higher level than Open Firmware (using a sort of fake System Folder on a small HFS partition - it's cool.)

  • "On a side note, how good is GCC on the PowerPC ?"

    IME - gcc sux on the PPC, it write very nasty unoptimized code. It works beutifully on x86, but not on PPC. egcs, OTOH, generates great PPC code so there is no lack of good compiles out there.

    I run LinuxPPC about 90% of the time, only switching to the Mac when I need to use the dreaded Microshaft Office - fortunatly thats rare. I only do this because my poor printer and Linux dont mix.
  • There's a rather brutal bug with varargs on the PowerPC Linuxes. C variable-length argument lists are implemented differently on the PPC platform, so code written for x86 will either not compile or crash mysteriously. I found this out the hard way compiling a SSH 2.0 client.
  • There are several other linux ppc distributions under way including:

    Debian Linux [debian.org]
    Rock Linux [rocklinux.org]
    [suse.com]
    Turbo Linux [turbolinux.com]
    APUS Linux for Power Up Amigas [sunsite.auc.dk]
    Yellow Dog Linux (Based on RedHat) [yellowdoglinux.com]
  • So far, almost all development has been with PCI based PowerMAC's. It seems everyone (except the benevolent folk's at MkLinux have forgotton that many PowerMac's feature NuBus. I know, it's slow, you say; proprietary you say, but by golly, when I've got 10 of these and want a beowulf... No, seriously, what else is going on to get Linux supported on the older PowerMAC's?
  • Statistically more people use Macintosh than use Linux. Historically and currently the average dollar per hour revenue created by employees using Macs is about 30% than those using Wintel. But what the hell, no one uses Macs, less use Linux, give up the fight and be assimilated. Is that your point?
  • by CoffeeNowDammit ( 5514 ) on Thursday April 20, 2000 @06:50AM (#1121773)

    One point everyone seems to be forgetting in the "my OS is bigger.. oops, better.. than your OS" debate is that PowerPC Linux has what may be the killer app.. the MacOS itself.

    Check out The Mac on Linux Page [maconlinux.com] for more info.

    I run it when I need to open that stray Word or PowerPoint document under Linux, and quite frankly, it rocks. Not everything is there yet (like sound support, and multiple serial line support), but what is there is impressive.

    (Aside: the Sheepshaver [sheepshaver.com] developers were supposed to have a similar app for Power PC Linux, but nothing ever materialized. Pity..)

    I'll soon try out additional Web resources for using MacOS Open Transport, MoL, and Linux ethertap support to create a "virtual" firewall. (Almost no configuration except for IP masquerading, no need to buy another box, and I get my MacOS-based VPN client when I need it. Life is good.)
    -----

  • by Valdrax ( 32670 ) on Thursday April 20, 2000 @04:37AM (#1121774)
    You probably disagree with all of those because you've never used it. Of course, you do lose the first 3 if you're Windows or X-fixated and can't approach the machine like a new user. If you give up all your assumptions about how a GUI should work, Mac OS is the easiest to train on. I do get frustrated as a Mac lab assistant at my college when someone panics about how to use a Mac because they're trying to do things the Windows way and can't step back an inch or two to look at the problem.

    Of course, your post was simply a troll, so I don't guess you really expected a calm answer, but, yes, the Mac OS has an interface that is all of the above.
  • by Valdrax ( 32670 ) on Thursday April 20, 2000 @05:02AM (#1121775)
    If BeOS actually cared about PPC Mac users, they'd hire people to do the same hard work that they LinuxPPC team is doing. The fact of the matter is that Be is targetting a higher marketshare with x86, and is using Apple as a scapegoat for their lack of continuing support for their original, most loyal customers. Plus, Jean-Louis Gassee gets to play the wounded victim of Big Bad Apple again, a part I'm sure he's loving.

    Besides, why are you slamming the efforts of a group that has done something you consider far too difficult for Be, Inc.? They're making Linux more available! What's wrong with this? Plus they're helping people not have to switch to a hardware platform many of us consider a shambling horror that just won't die, and they're helping wring extra life out of old machines that might be wasted otherwise.

    Shouldn't we be supporting Linux everywhere? I mean, everyone loves it when Linux is ported to a digital camera, a calculator, or a friggin' vending machine. What's so wrong with giving Mac and other PPC users a chance at Linux without having to give up their current machine or get an extra one? Quit slamming people for providing others with a service.
  • by hey! ( 33014 ) on Thursday April 20, 2000 @09:59AM (#1121776) Homepage Journal
    OK, this is off topic, and its been a while since I've done serious work on a Macintosh.

    It's clear that in areas of memory and task management, MacOS is very weak.

    However it does have areas of strength too.

    For example Applescript (is it in or out this year) is dynamite. It allows you glue together Mac applications the way the Unix shell allows you to glue together command line applications. It also gives you the ability to expose the objects your application handles to the outside world in an OS-wide standard way.

    I like the way the OS manages application resources. Essentially, strings, menus, bitmaps and so on aren't statically compiled into applications or into dlls, but are kept in a separate "fork" of the application file which can be edited with a resource editor by the user.

    The UI is generally quite good, although apple has been slipping in recent years. I like the fact that the OS has a user notification API. This is the most annoying thing about windows: having applications pop windows to the top of the Z-order and steal keystrokes.

    The UI is good in many subtle ways too. For example, the way menus stick to the top of the screen. It may seem weird at first, but consider this: the ease of hitting a target with a on screen pointer is dependent on the size of the object (obviously). Not obviously, by sticking the menus to the top of the screen they're given infinite vertical extent making them easier to hit. Now, for the average, non impaired person, person, this is not a huge thing; it saves you a tiny bit of effort.

    The earliest decisions made in the development of the Mac GUI were often like this -- very astute. Enough right decisions, subtle though their effect may be, and the whole user experience is improved.
  • by imac.usr ( 58845 ) on Thursday April 20, 2000 @04:38AM (#1121777) Homepage
    A plea to Linux hackers everywhere - if you're coding on an Intel platform, do the best you can to keep x86-specific things out of your code, so that other Linux/BSD users out there can enjoy the fruits of your labors.

    Remember, there's PPC, SPARC, MIPS, Alpha, and 680x0 ports out there as well. Are you listening, Corel?

  • by dlc ( 41988 ) <dlc.sevenroot@org> on Thursday April 20, 2000 @04:29AM (#1121778) Homepage

    I know I shouldn't respond to this obvious troll, but here I go...

    • Does anyone use the Mac as a serious computing platform any more?

    What does that mean? What is serious computing? There are many univesities that use Macs for teaching. There are many companies that still have a lot of Macs, especially for graphic design. There are many households who are still using Macs, and surprise, surprise, the marketing is working, there are many households that are actually buying Macs. Besides, a telnet client is a telnet client is a telnet client, right? :)

    But just to reiterate -- this is not about Macs! This is about the PowerPC architecture. The hardware was never the weak link in Macs -- it was always the software. The hardware has always done an excellent job running the slow, underpowered, under-equipped software.

    • top of the range PC will still beat it hands down for performace

    Actually, this is not true, contrary to the reports you might have seen. Yeah, a truly high end PIII or Athlon might beat a G4 for raw performace, but the beauty is, G3's and G4's are common, and relatively inexpensive, while the high end x86 chips are rare and expensive. The average G3 will toast the shit out of the average PII or PIII.

    darren


    Cthulhu for President! [cthulhu.org]
  • by fantomas ( 94850 ) on Thursday April 20, 2000 @04:40AM (#1121779)

    There's an awful lot of loyalty in arts and design community towards the Mac.

    Windows boxes are viewed by this community with a similar sort of suspicion that the linux / slashdot community holds for Microsoft products, though this is more due to poor interfaces, bad ports of software and perceived complexity of operation compared to the familiar Mac environment, than any issue of open or closed source software. In the UK art colleges are often strongholds of Macs, this is what most artists and graphic designers are professionally trained on.

    However, there are a lot of graphic designers out there who are becoming aware of the whole open source philosophy and would love to join in. To drag up the old chestnut, they are just waiting for a decent set of tools that are suitable for their work environment. I work with a graphic designer who would love to convert to linux but the tools available are just not comparable to the commercially released latest versions of Illustrator, Flash, Quark Xpress etc.

    With the linux heritage and community developing from people interested in unix server applications to including those interested in providing graphics and multimedia tools, I think we may be on the verge of seeing a new professional community embracing linux and open source.

    Ok, I know it's very easy to say "forget the damn graphic designers, if they can't learn our way of working, we don't want to help them" but I think if the attitude is more "let's work out how to support these user needs" we'll find a large group of unsupported sympathisers becoming enthusiastic converts.

  • by MoNickels ( 1700 ) on Thursday April 20, 2000 @06:38AM (#1121780) Homepage
    I'm a stone cold geek! I got that hacker instinct running around in my thick Mac head, and I got your computer right here, buddy. It's an Apple PowerBook G3 running PowerPC Linux. Whooop!

    I run Virtual PC just so I can rape and pillage in Windows 98 and Windows NT 4.0 client, as well as Mac OS 9 and PowerPC Linux. I mate them like dogs in heat in July. I do it for fun, not cause I have to. I can play Tetris on every one of them, and I my high scores show up at the bank. Duke Nukem licked my boots. I'm about to add a RedHat partition, and I've got MachTen in a box on a shelf just panting to join the gangbang.

    I flip around from OS to OS like a Vegas master dealing blackjack. I do what I want where I want any way I please. I compose a document using AbiSoft Word, or ApplixWare, or Microsoft Word, or ClarisWorks, or in VI (God bless the simple things in life). I've got 17 graphics programs, 18 web browsers, 41 apps capable of serving web pages, 10 text editors, 7 shells, and one goddam processor. And the machine weighs seven pounds. I can send email from 19 different programs, and just for the hell of it, I can first bounce it through 14 accounts on three planets. I can read your reply, translate it into French and post it on the web in Chinese. My computer sings to me.

    I can jam my baby, my rocket, in the middle of two ten-thousand-node networks and have it act as a router, a bridge, an end-node or a firewall. My machine does IP masquerading like a Halloween ball. I can grep like a mofo, find just the true-life Pantone color you're looking for, and visit your dirtbag Windows-only web site just so I can send you nasty email and jam your mailbox with stories of the Craig Shergold and free trips to Disneyland. I eat Lithium Ion batteries for breakfast and chew NiCad cells to calm my stomach.

    My machine is the best of breed. It's got hybrid vigor. You can take your fancy-pants, out-of-the-box, turn key solutions and spread them on the grass, 'cause they ain't nothing but manure.

    [With thanks and respect to Mike Fink and Mark Twain.]
  • by um... Lucas ( 13147 ) on Thursday April 20, 2000 @04:48AM (#1121781) Homepage Journal
    The Macintosh still kicks Windows' butt in the publishing arena... So many of the applications over the years have come to the Macintosh first and then windows, to the point that only recently have applications been released simultaneously and with feature parity across the two platforms. Apple has also integrated many things into their operating system that aids publishers... ColorSync (where is Windows' color management), QuickTime (if an app supports quicktime, then it supports all of the file formats that Quicktime does with zero extra effort from the developer).

    Many designers learned their trade on the Macintosh these days... Many of the old school ones converted to Macs a long time ago... The MacOS might not be as stable as the Windows platforms (which I'd debate, being a user of both), but it's MUCH easier to setup and maintain, allowing non-technical people to go about being creative, rather than caring about DLL's, the many security flaws in Windows, etc...

    Besides that, in case you hadn't noticed, Apple has doubled their market share in the past couple of years... Meaning their market's growing. More macs are being sold, not less... so why do you insist on why should so and so be ported to the Mac?

    I do understand your second sentiment a little bit though... Mac users love their desktops, so why would they want to switch to Linux? And then the next question would really be: Why would a mac user want to switch to Linux with OS-X right around the corner? That's a real toughie, I think...

    The first answer would be that some mac users are actually curious about other technologies...They use a Mac because they feel it's superior to Windows, but they're not afraid to look at other options available. The second answer? I really don't know... As i said earlier i'm a long time Mac user, though I've also PC with Windows and Linux on it, and i really do have zero interest in installing Linux on a Mac these days... A year ago, I played with MkLinux, but now that OS-X is just about here, i'd rather sit back and wait for that.
  • by hey! ( 33014 ) on Thursday April 20, 2000 @06:06AM (#1121782) Homepage Journal
    My boss is a mac fanatic; he can no longer work exclusively on Macs because none of our regular customers use them, so he works pretty much bouncing betwen the Mac and a PC. One side effect of this is that new Macs mysteriously show up in our office, because of a rare customer who orders one and changes his mind and the bossman can't bear to part with a new Mac.

    So we had a blue and white G3/350 sitting around as a doorstop. My knee jerk reaction to any doorstop computer is to put Linux on it and put it to work! So I picked up Yellow Dog Linux, which is Red Hat 6 recompiled for the PPC and tweaked.

    The first big difference from x86 Linux -- no PC BIOS, therefore no LILO.

    The problem is that it turns out is that "Open Firmware" isn't. It is supposedly possible to boot Linux directly on powerup, but it is apparently enough of a black an art the normal course is to boot Mac and use a Mac Control Panel applet called BootX to automatically load Linux. YDL does not include any instructions for booting any other way. I wonder what will happen when OS X starts shipping on new Macs -- until the Open Firmware problem is solved, you'll have to have a Mac OS CD to install a small HFS partition with a stripped down MacOS 8 or 9. This isn't really all that bad when you think about it. Think of the HFS partition as the boot sector of the disk and MacOS as a GUI boot sector manager; heck you even get networking and a web browser thrown in.

    There was plenty of hassle getting the G3 to boot. I resolved this after some Internet searches to find the magical incantations needed by BootX specifically for use on the Blue and White G3. Also, it may be necessary to tweak some BootX settings so you don't boot with the processor cache turned off and the video hardware set to VGA resolution so RTFM. After ritually sprinkling the keyboard with virtual chicken blood, I booted into the familiar RedHat installer. From there installation is pretty much coasting.

    Once you get over this stuff, what you end up with is essentially identical to Red Hat. It's shipped with slightly more paranoid security defaults, which is a good thing. Instead of wuftpd, you get proftpd, which is a nifty ftp daemon that uses Apache-like XML files for configuration. Unfortunately turns out to be mysteriously unstable on YDL; it would run fine for a while suddenly start losing its mind and not allowing any logins. I downgraded to wuftpd and all is well. Also, GNOME is unstable and dumps core after a little while, but KDE works without a hitch. I wonder if these problems could be some of the compiler glitches mentioned in other threads. In any case, combine this with the annoyance of the round, one button mouse (hint, use the keyboard "=" for the right button mouse), and I wouldn't really recommend this solution for desktop use.

    Pretty much these are minor issues, and the thing works fine as a server. I got Apache and MySQL running on it in a few minutes. I got the latest Python and recompiled with threads to run Zope, and multihomed the Zope with a reverse proxy Squid. I obtained OpenSSH 1.2.2 and successfully compiled it (I've heard reports of problems with OpenSSH 2 on PPC linux). The point of which is that all the usual open source tools are readily available, but that a few have some glitches.

    Performance-wise, the system is nothing to write home about. I have a P3-450 which kicks its ass readily on long compiles and on Zope service. I'd say its performance runs something between a PPro 200 that I have and the P3-450. However, performance is perfectly satisfactory for a former doorstop.

    I suspect the glitches and relative sluggishness may be related to the fact that the PPC compiler is less robust than it's X86 counterpart.

    Bottom line: YDL Linux is great for repurposing existing Mac PPC Hardware (but not too old and probably not too new) for use as a server in non-demanding missions.

The difficult we do today; the impossible takes a little longer.

Working...