Slashdot Mirror


Darwin on Crusoe?

MacOS Rumors is running a blurb that Apple is exploring porting Darwin to other processors (including Transmeta's Crusoe processor) due to frustration with availability of high speed Motorola G4 processors. An interesting though, a Mac without a Motorola chip... Of course, it's just a rumor at this point... (update: I've got it confirmed from "anonymous sources" that this is true)

15 of 257 comments (clear)

  1. Why not just use the Crusoe as a G4? by Frater+219 · · Score: 4

    Instead of "porting" Darwin to x86 and running it on a bog-standard Crusoe, why not talk the Crusoe into being a G4 instead of an x86? I was under the impression that the Crusoe's code-morphing software could be reworked to emulate other instruction sets besides x86 ...

    1. Re:Why not just use the Crusoe as a G4? by jovlinger · · Score: 3

      Hrm. Don't agree with you there. Well, I can't argue the facts of RISC and CISC, but...

      The main thing about risc is: each instruction takes one cycle to complete. This has secondary implications on the instruction being simple, but the whole point is the uniform length. This is used in:

      Superscalar processors have multiple issue for risc; the hardware necessary to translate the sequential instruction stream coming in to an equivalent 4 way (optimally) parallel instruction streams is VERY hairy. Note that this is really only possible if we know that each instruction is the same size (completes in the same number of cycles).

      Note also that ALL modern x86 processors translate x86 to some internal risc code, and then use exactly this technique to keep those functional units busy.

      However, Crusoe has VLIW instructions, which basically are X way parallel instructions. The mophing software does the superscalar issue statically (at translation time) rather than at issue time, but still; crusoe is a multiscalar RISC cpu (+ other cruft). When Crusoe morphs x86, it first riscifies it, and then schedules it.
      So it is easier for crusoe to translate RISC instructions, as they are already RISCified.

      Now, it is hard to do as good a job of issuing instructions statically as you can do dynamically, but it does simplify the chip a lot, which I guess you can use to make it less power hungry.

      Johan

    2. Re:Why not just use the Crusoe as a G4? by chadmulligan · · Score: 3
      The Crusoe might well be a good emulator for the generic PowerPC architecture, but I doubt that it will be able to get any comparable performance emulating an AltiVec unit... and much of the G3/G4's non-AltiVec performance is due to the excellent backside cache implementation. I don't have Crusoe's specs handy at this moment, but their cache architecture will necessarily be less potent because of power constraints.

      Now, a Crusoe implementation of PowerPC for a hand-held - a Color Palm running some subset of Mac OS X - that might be both feasible and interesting. Regarding other achitectures, porting has always been done at Apple for internal projects - starting with the defunct "StarTrek" port of System 7 to the x86 - but there's been talk for years about this or that port, with no product emerging. Now they're talking about a a modified AMD chip; but remember, there's a very heavy investment in PowerPC binaries, and PowerPC-optimizing compilers. Even if it's just a simple recompile (which it never is), you're talking about major manpower involvement...

      MacOSRumors always makes for interesting reading, but it's just rumors, don't forget.

    3. Re:Why not just use the Crusoe as a G4? by retep · · Score: 4

      That's right. CISC chips depend on a small number of big instructions running relitivly slowly. RISC chips depend on lots of little instructions running super fast. The Crusoe can't translate RISC instructions efficiently because there are so many of them all at once. So RISC gets pretty bad performence compared to CISC.

      This doesn't just apply to the Crusoe. Any emulator has the same problem.

    4. Re:Why not just use the Crusoe as a G4? by richnut · · Score: 3

      The thing is, nothing needs to be "ported"

      Darwin exists for x86 (in some form or another) and always has. All of that code was x86 back when it was Openstep, (and 68k when it was NEXTSTEP). Rhapsody Developer was on x86, and the early versions of Darwin (cant speak of recent ones as I have waning interest) also included x86 source trees. I never had much luck building the first release of Darwin on Openstep 4.2, but I dont doubt it was possible for the truly motivated. Apple could already have code running on Crusoe if they wanted since it's likely they still have some of that expertise around.

      -Rich

    5. Re:Why not just use the Crusoe as a G4? by um...+Lucas · · Score: 4

      I would think that's what they're talking about, being that otherwise it could just be said that Apple's considering porting their software to run on x86 processors. That says nothing about all the support chips. Apple will never have their OS run on BX motherboards, as it would significantly detract from the users experience of "trouble-free hardware".

      IF they could just modify the pinouts of the Crusoe to conform with their sockets... They'ed have a pretty cool setup.

      They'ed also probably want to emulate a G3 rather than G4...Somehow I doubt that Crusoe could emulate AltiVec very well... Or maybe there is real magic to Crusoe.

      Anyways, it'd be exciting to see Crusoe emulate ANYTHING but x86... Right now it just seems like it's a low power x86 processor. Not very exciting.

  2. Read a real G4 report by ryanmeader · · Score: 3
    Don't you understand that i make this shit up?!?! Everything i have on my site is complete bullshit that i make up! I wake up at 12:30, take a hit off a bong, and head over to the 6500/250 that someone donated to me, and i invent "stories" from info i read from real websites.

    That being said - i was with apple sticking with the PPC for a long time.. but now, with almost a _year_ of failure... i'm beginning to believe that there needs to be another way to power our Macs. We have almost 12 months of proof that Motorola can't cut it. Hell, they care 20 times more about embedded processors - and are simply not _that_ interested in high end PPCs.

    Now, if you want some _honest_ reporting of what's going on, besides the bong resin i spewed out on my page, go read about the G4 mess over at AppleInsider. It tells about the problems that Apple is having with Mot and how they went begging to IBM to make AltiVec enabled PPCs.

    I do think that Apple is in real trouble with the PPC series.. and its not the technology, its the companies involved.
    ______

    --
    ______
    Black light Media is looking for some good advertisers!
  3. Mac on IA64?? by josepha48 · · Score: 3
    It would be interesting to see a Mac OS running on an intel chip like the upcoming IA64. Especially if it ran all the MS products that Mac currently does. Can you imagine what would happen to windows? I'd be intereseted in a Mac port to other processors, I think that would make thie indulstry rather interesting again. I'd also like to see Darwin and how it would perform on Intel against other OSes like Windows, Solaris, and Linux on the intel platform.

    It woudl be even more interesting to see Mac on a Sparc chip or an 1000Mhz Alpha. Now that would be speed in style. Think of the graphic capabilities that would have.

    send flames > /dev/null

    --

    Only 'flamers' flame!

  4. Re:MacOS Rumors: Not Very Credible by Phroggy · · Score: 3
    In MOSR's defense, I've been reading the site pretty much daily for a very long time (let's just say I remember when the URL ended in /~rumors/), and I've been quite impressed with it. They've never claimed to be more than a rumor site; in fact on numerous occasions they've reminded their readers that that's all it is.

    Sure, they get misinformation sometimes, but they don't misrepresent it as fact, and the correct information far outweighs the errors.

    It's conceivable that Apple could work with Transmeta to make a PowerPC clone. But that's what it would have to be - same architecture, same instruction set, different chip.

    I remember thinking awhile back that AMD would make an interesting addition to the Apple-IBM-Motorola partnership - after all, AMD already licences technology from Motorola, and vice-versa I think.

    <!--#include file=".signature" -->

    --
    $x='S24;r)>63/* h@<5+oZ)32"5cz';$me='phroggy'x$];
    $x=~y+ -xz+\0-Tx+;print$_^chop$me for split'',$x;
  5. Re:You misunderstand Apple by um...+Lucas · · Score: 3

    But that wouldn't be a Mac. It'd be a bastardized Mac-like implentation. No matter how much you try to justify it, it they just no longer would be the Mac's that we've known. A lot of the Mac OS' ease of use comes from it's closeness to the hardware. Things like Plug and Play (which apple never bothered to name as such) work much better on every iteration of mac compared to PC's. They also do USB much better than PC's. That's not just beacuse of a couple extensions to the Mac OS. It's beacause Apple knows EXACTLY what chips to expect in each Mac system, as well as tolerances of things like motherboards and cables. Microsoft can have pretty much no assurances that vendor A's USB implentation is anything like vendor B's (especially so the smaller the vendor is)... they may even use the same chipset, but place them at different distances from the PCI brideg chips, messing up the timing...

  6. Re:Apple..... by mstone · · Score: 3
    > My point is that Apple is missing a fairly large market share...
    >the roll your own crowd.

    that market isn't as attractive as you might think. it's composed largely of people who want to buy components on the cheap. the margins there are thin, and strongly susceptible to economies of scale. if you're not one of the really big fish in that market, or one of the hyerspecialized little fish, you're fish food. that's why we're unlikely ever to see a vertically integrated hardware/OS/software vendor in the x86 clone market.

    Apple's marketing is strongly based on transparent hardware support and ease of use.. "it works, right out of the box", "this is the manual you get with a Mac", "there is no step three", etc. that requires tight synchronization between hardware and software, and the roll your own crowd could traditionally care less about that. they want to build what's cheap, not bow down before someone's list of officially approved hardware. the last thing Apple needs is to have its brand shot down by a proliferation of nightmare boxes built by people who don't give a damn about Apple specs.

    besides, it costs like hell to support umpty-seven different permutations of hardware at all, let alone reliably. Microsoft has trouble doing it, and they're just a tad bigger than Apple is. you even hear occasional stories about Linux users not being able to get a certain hardware configuration to work, and we have a fairly decent amount of support, too.

    so when you take the various factors into account.. loss of margins on the hardware side, negative branding from noncompliant hardware in the marketplace, and a geometric increase in the cost of support on the software side.. you'd have to buy a whole lot of Apple-produced software to offset what you'd cost them.

    just between you, me, and the fencepost error, i don't think you were planning to spend more than a couple $K on software, were you?

  7. MacOS hardware and software problems and thoughts. by chainsaw1 · · Score: 3

    While this Would Be A Good Thing, it's not quite this simple. The motherboard would have to include the Apple ROM, which is one of the primary keys to getting MacOS to run on other hardware archs. There is also the matter of the device support. MacOS MoBoards have so much on them: 16bit sound, Firewire, etc. old Macs has SCSI-1, video, etc. on the board also. Finally, the BIOS for MacOS is their own product called OpenFirmware, which they would have to either write around or release compatability info on.

    Now don't get me wrong, the MacOS hardware software dependency is going towards something where this may be possible. For instance:

    The ROM is slowly being moved into a system file which gets loaded into RAM. MacOSX doesn't (read: "shouldn't") access the ROM at all, I believe (Yay!)

    Video and SCSI are now off the board. Video is going to be standard AGP (as in Sawtooth G4's) rather than on-board or in a special 66MHz yet 32 bit PCI slot.

    MacOSX early versions/Rhapsody Dr's were released under certain x86 platforms already. MOSR had a report of someone seeing a Rhapsody DR at Cupertino(sp?) running on a Sun SPARc 4.

    So there are hurdles to overcome, but Apple has been overcoming them slowly anyway. The most likely thing that I would see happening is Apple making motherboards based on their hardware for different CPU (AMD-Motorola alliance crossed with Apple-Moto-IBM's 'AIM' alliance may yield something interesting...K7's maybe?) would probably be the most likely outcome.

    In this case, you will still have a premade system from Apple then build on it.

    --
    - Sig
  8. Re:correction on Open Firmware by Anonymous Coward · · Score: 5
    Open Firmware isn't unique to Apple. It's actually an IEEE standard for boot firmware. (The "Open" in its name really does mean something.) From what I remember, it's a standard created by Apple and Sun, evolved from Sun's earlier firmware (called OpenBoot). It's actually pretty cool. It has a Forth interpreter built in.

    On Macintoshes, you can access it by holding down Command-Option-O-F while booting. On Suns, it's Stop-A. I believe someone actually wrote a Forth version of Breakout that could run on Apple's OF implementation. Pretty neat trick for a boot monitor, imho.

    Some PCI devices even have Forth driver code in their ROMs.

  9. Re:Apple Might (Not) Be Awakening by Mononoke · · Score: 4
    Apple Computer Corp. has such a history of screwups, bad decisions, bad marketing, etc. et. al. that I have written them off.

    Translation: "I got scared during the dark days when Apple was 'beleagered,' so I ran and hid my head in the sand."

    I just read this morning how they chose not to give or even sell (at street price) iMacs to freeiMac. The company FreeiMac would give away free iMac computers with the usual 3 years of Earthlink subscription plus accepting advertising.

    Translation: "I read somewhere that Apple didn't want to supply tech support to thousands of users too dumb to know that there's no such thing as a free computer, and anyway, Apple already has an Earthlink deal going."

    Talk about clueless! They have turned down another opportunity to increase the awareness and use of their computers. They already have an ever decreasing share of the market and decisions like this will only help seal their doom.

    Translation: "What is this iMac thing everyone is talking about? Why have I never heard of it? I'm sure glad I sold all my Apple stock when they reached that high of 18."

    Jobs is an insane, megalomanial leader of a corporation led by more fools. Not that Apple doesn't do good things -- the Macintosh is an incredible computer and I loved mine. But I bowed to economics and bought a less expensive PC (I prefer Linux or BeOS, thank you).

    Translation: I've heard that Steve Jobs runs a tight ship, and forces his employees to work towards a common goal, and to be productive. It must suck to work where you have to earn your paycheck. Since I don't have a paycheck to worry about, I couldn't afford anything but this leftover PII doorstop, and whatever free OS I could find to run on it."

    This possible move to the Crusoe processor may be the sign of intelligence: Emulating the G4 in a cheaper and less power-hungry processor == Less expensive G4 == Cheaper Macs == More consumers buying your expensive hardware.

    Translation: "I don't know much about the G4, but I assume it uses electricity just as inefficiently as a Pentium. I wonder if I can get a free Crusoe. Too bad I flunked economics."


    --

    --
    NetInfo connection failed for server 127.0.0.1/local
  10. MacOS Rumors: Not Very Credible by MoNickels · · Score: 5

    I don't know if the porting story is true, but Mac OS Rumors has a spotty record. Many of its leads often seem to be lifted from elsewhere, the rumor reliability is poor, and usually so late as to be useless, besides which its articles are so filled with bogus predictions using words such as "likely", "possibly", "predicted", and longs chains of if-then scenarios as to completely obliterate any authoritative source that might have passed a true lead on in the first place.

    You will be issued one grain of salt apiece.

    --

    Wordnik, a dictionary project which aims to collect