Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Apple Businesses

Apple Moving To G5s Next Year? 168

Posted by CmdrTaco from the rumors-and-speculation dept.
Rand Race writes: "Lowend Mac is reporting here that Apple is considering dropping the G4 on most of its line in favor of IBM's Altivec-less G5. The G5 would appear in early 2001 at speeds from 700Mhz to 1Ghz in single, dual, and quad setups. Speeds up to 2Ghz should be reached by 2002. Updated G3s (750Cx/G3e) at 533-666Mhz will be used in iMacs and iBooks (maybe a 666Mhz G5 iMac SE), Dual and Quad G4-500s in midrange machines, and G5s in high end machines (733,866, & 1Ghz), Powerbooks (733 & 850), and Cubes (866 & 1Ghz). Disclaimer; this is a rumour and we all know about Apple rumors..." Update: 08/18 09:04 PM by CT : Several people noted that this is more then a rumor, its a blatant lie... they got very worked up about it too ;)
This discussion has been archived. No new comments can be posted.

Apple Moving to G5s Next Year? More

Apple Moving to G5s Next Year?

Comments Filter:
  • Have you ever seen a program does REQUIRED MMX, SSE, 3DNow? The truth is aside from Photoshop and some other special applications, no one uses Altivec that much because they want to remain compatible with G3s. Having the G5 (which is used in IBM workstations and servers) gives Apple an edge in server applications where Altivec isn't really that useful.
  • Now that I think about it, you would have noticed the troll if you had read the fine print:

    The Rumor Mill is part of Low End Mac. This site has no affiliation, actual or implied, with Apple Computer, Inc.
    The site exists to provide an alternative to real rumor sites. Any logos, slogans, names, or representations used or made are done so in the above context. All information on The Rumor Mill (and the rest of Low End Mac) is copyright ©1999-2000 by Daniel Knight, unless otherwise noted, and may not be reproduced in any form without prior consent. All Rights Reserved.

    (Emphasis added by me)

  • Makes me wonder if Motorola could/would offer an altivec unit as a standalone coprocessor.


    blessings,

  • Comment removed based on user account deletion

  • sigh... please read the pages you submit. (Score:5)

    by option8 ( 16509 ) on Thursday August 17, 2000 @08:47AM (#848832) Homepage
    the rumormill on lowend mac is not even a real rumor page, but is intended as humor - a jab at all the 'real' rumor pages (macosrumors, appleinsider, etc)

    have a look at http://lowendmac.com/rumormill/about.shtml:

    Okay, for those of you unsure about this site, here's the lowdown: it's a farce. There are no contacts at Apple Computer. There are no leaks from anyone reaching us. There is no Anne Onymus.

    It's all made up. The Rumor Mill is intended to parody Mac rumor sites, although at times it may be hard to tell. (That's why I posted this page.)

    I really hoped I wouldn't have to say that. I really hoped the broad-faced lies would be pretty obvious. I really wanted it to stand as thought-provoking parody.

    But then it struck me, what if I guessed something right. Apple Legal would be on me like hunny on Pooh! So, here's the page that says it's just for fun.

    As for the other rumor sites, they seem to take themselves a bit more seriously.

    As for who Anne Onymus really is -- that would be telling.

    - Dan Knight, publisher, Low End Mac

    i hate it when i see something like this, but there's already a hundred posts going on about where are the numbers? and who is the source?

  • I've heard they've got OSX running on Alphas and have heard very faint rumors of them working on Crusoes.

  • Re:There is no G4 chip (Score:1)

    by Anonymous Coward
    The G4 is both a chip AND a computer system.

    Don't correct people if you don't know what the fuck you are talking about.


  • Oh, the G4 has really given Apple plenty of marketting trouble. People just love that megahertz number despite the fact that it is only one part of a fast machine. Apple has done a good job showing how the G4 boosted Photoshop performance, but I don't know if it's helped much else. Is there a more general set of benchmarks? Is there any other application that's sped up by the Altivec? Does anyone ever use Intel's MMX instructions?

    I remember when we had MIPS, a reasonably good standard. Are any on the horizon?
  • Having the G5 (which is used in IBM workstations and servers) gives

    As far as I know there is no such thing as a 'G5' used in IBM server and workstations. There is the POWER series of processors which are power hungry 64-bit processors and have very little to do with the low power embedded style processors that motorola produces. Look at RS/6000 Model F80 if you don't believe me.

  • Gotcha! (Score:5)

    by Anne Onymus ( 223599 ) on Thursday August 17, 2000 @09:00AM (#848837) Homepage
    Good afternoon, Slashdot, and thanks for liking to my delightful PowerPC G5 rumor hoax. If you'd read the whole article, seen my other articles, and read the Rumor Mill about page, maybe you wouldn't have linked. OTOH, as my publisher notes, it's remarkably easy to create a plausible rumor with absolutely no basis in reality. I've loved your discussion, though, especially the 666 observations. I'll have to put in for a bonus after getting Slashdotted. ;-)
  • IBM demonstrated a PowerPC chip running at 1 Ghz way back in 1998

    If you look a little more carefully at those press releases you will discover that chip would never have worked in any real system considering it acccually a implemented a subset of the required PPC instructions. It was a technology demo nothing else. IBM's up and comming POWER4 on the other hand is a real ghz processor but apple probably won't want to ship a processor designed for a server in the near future. If apple wanted 64-bit PPC's with fast memory subsystems they could have had them. Of course apple will have to give up that we don't need a fan attitude when their processors are designed for high power computing tasks.
  • n all, its really a shame that this fiasco has occured. The PPC has a much cleaner design than the x86 based chips.

    Are you refering to the actual silicon or to the instruction sets? there is a difference, wish i could remember the URL. The actual chip designs have nothing to ddo with the instruction set in the x86 line, however it has everything to do with the ppc line. since ppc's don't have a "micro core" to translate instructions to something that the processor actual uses, therefore x86 based chips can under go radical redesigns more easily the the ppc, IMHO this is causing the Mhz problems.

  • But MMX, SSE, and 3DNow! are hard to use in the first place. All that sharing of registers and register switching crap cripples the performance of SIMD. There's a tradeoff involved, and you may end up reducing performance if you aren't careful. Altivec involves a completely independant register set, so you can only really gain in performance. Of course, this is all a theoretical point if Motorola can't produce the chips with decent yield. And they certainly aren't!

    I don't understand why there is so much trouble with Altivec. The UltraSPARC, HP PA-RISC, Alpha, and MIPS processors all have SIMD units. Even the cheap embedded processors like the SH4 have integer and fp SIMD. I think Motorola was actually the last to add support. I doubt there's an inherent problem with Altivec. It's probably that Motorola's G4 design doesn't fabricate well.

    Motorola has had a lot of trouble in the past with their designs. Many of their 64 bit designs (the 620 and 640) were scrapped because of manufacturing problems. I believe that the only surviving member, the 630, is what IBM calls the Power 4.

    BTW, IBM doesn't use the G5 at all right now. It doesn't really exist yet. They use the Power 4. The G5 is Motorola's new 64 bit design.
  • You stupid fuck, you claim not to care about Apple rumors but here you are.

    Dude, I think you read it wrong, or I wrote it wrong. What I'm asking is why Apple makes such a fuss over squelching rumours, not why /. makes such a fuss over rumours.

    I, for one, love reading about the rumours. They make me want to see and buy the product even more. What I'm saying is Apple should (and probably does) know this and probably works it to build the hype.

    --Calum

  • Re:apple's problem... (Score:3)

    by pi radians ( 170660 ) on Thursday August 17, 2000 @07:37AM (#848842)
    What are you talking about?? The G4's (other than the cube) all come with PCI slots, a 2x AGP slot and firewire.... plus room to add new internal components...

    and speed is their biggest problem... they've been held at 500 MHz for almost a year now.

  • seriously, though -- does anybody know if the new dual processor macs will run linux? isn't that the real question on everybody's mind?
  • Hasn't Jobs sued you yet?
  • that space above the DVD drive is not a drive bay, even though it may appear to look like one when you open the case. As there's no corresponding hole at the front, so the olny thing that can be put there is a hard drive

    Correct. It was originally for a single HD. The space *below* the DVD bay is for a 3 1/2" removeable device.
  • Whoops, correction: the PowerPC 630 actually became the Power 3, I believe. The Power 4 is a new thing.

  • And (Score:1)

    by sulli ( 195030 )
    Apple Sues Cobalt Over Cube" [lowendmac.com], and others. These guys are doubly funny now that /. is taking them seriously!

    sulli

  • >But looking at the Wintel market, where Altivec and 128 bit register are unheard off, Speech Recognition software is much further ahead of anything available on the Mac

    True, but I think that has more to do with the historical apple-version-lag than the usefulness of Altivec in speech recognition.
  • Mac gaming is far from dead. It is true that the raw quantity of games is much higher on the PC side, but the majority of "big" games are available for the Mac (Unreal Tournament, Quake 1/2/3, Diablo 1/2, Deus Ex, etc.) If gaming is your primary purpose for buying a computer then don't get a Mac, but if it's a secondary purpose you'll probably find that there are many more Mac games that you want than you have time to play. At least that's the case with me.

    BTW, Bungie did get sucked up by MS, but they have announced that Halo will be a simultaneous Mac/XBox release.

  • Why'd they put a less-capable, albeit faster, processor in a HIGHER end machine?

  • It's not a gimmick. SIMD worked pretty darn well on the Cray, and is heavily used in DSPs in the form of multiplier-accumulators. The only problem is that you need code that actually uses it. It works great for matrix multiplication. You can tell a processor to multiple a row of a matrix by a constant, and it can then perform it as a parallel operation. Without it, you'd have to hope that the processor realizes that it can perform the multiplications in parallel. And image processing and 3D graphics software do a lot of matrix multiplications.

    The reason we don't see much benefit on Intel processors is because the implementation is seriously limited. MMX shares the same registers as the FPU because expanding the register set forces all operating systems to be updated to save extra registers on a context switch. Microsoft would have had to release new versions of Windows, and Intel didn't want to go that route. And the x86 FPU unit is stack based, which makes it hard to optimize in the first place. The only decent implementations were on processors with little software support to begin with, except for the G4. The only evidence we have of SIMD's advantages on the desktop is from Mac OS X. But that isn't released yet, so we don't all have first hand evidence.

    And the reason that someone is backing away is because they can't manufacture their chip properly. In a situation like that, a company would rather say the feature is not that important than say their design was a failure.
  • The PPC chip is actually an incredibly excelent chip, it it works well. The problem is Motorola who really doesn't know where it is going and even has conflicting interests. On the one hand the produce the PPC processor and the other hand they do very little to push it and develop mother boards for the Intel and Sparc chips. Its strange that Motorola is incapable of seeing a good product when they see it.

    Whether the story is for real or not, shouldn't really be the issue, what should be examined is whether such a move would make sense. While on the one hand the altivec does stuff differently from the MMX, IBM has probably got it right the KISS (Keep it simple stupid) approach as it simplifies on production and helps them concentrate on other aspects of the chip production. After all we are talking about a RISC chip, where the whole point was keeping out junk - just see where CISC has got Intel. Although the x86 line is still advancing, Intel realizes that at some point they are going to have to remove a lot of redundent instructions if they are going to reduce development times.

    If Apple chucks altivec, and goes to someone whoe believes in the PPC, then I will be happy. Also since IBM is producing the PPC in large quantities, Apple may even be able to get some cheaper chips.
  • Your = possessive
    you're = contraction of "you" and "are"

    And it's "grammar."

    Dumbasses. Finish "grammer" school before you post next time.
    --
  • With that aside what ARE the alternative CPU options for Apple?

    This is a completely serious question because I honestly do not know much about Crusoe. Would a Transmeta processor be plausible here, being that their architecture is supposed to emulate other processors well?

    Apple is using it in it's consumer boxes, IBM in servers, & Motorola is suplying Apple while focusing on the embedded market.

    BTW: IBM also has been manufacturing PowerPCS for Apple for the past several months, and is apparently doing a much better job of it than Motorola.

    - Scott
    ------
    Scott Stevenson
  • Steve Jobs gets to demonstrate what a big crybaby he is.
    ----------
  • Meanwhile, PPC can trash 1.1GHz x86s in some tests... So can 600MHz Alphas. Clock speed isn't everything.
  • There's also a crazy rumour that MacOS X is up and running on Alphas. {shrug}
  • Well, this is referring to the IBM G5 chip. IBM and Motorola actually differ on their processor numbers even though they use the same name. The G5 produced by IBM will actually be a variant based on the Power4 (IBM G4) architecture. Motorola's G5 will be a multi-core architecture based around the Motorola G4.

    AltiVec is a technology developed by Motorola. It's primary usage is for accelerated processing of stream data. It performs the same operation over and over on a stream of data. The main purpose for Motorola's development of AltiVec was to be of use in the embedded market, which is their primary cash source. Since IBM focuses mostly on the high-end market such as servers, this wouldn't be as useful for them. The specialized acceleration provided by AltiVec wouldn't provide as much benefit as raw generic processing power would.

    It's somewhat unlikely that Apple would move to the IBM G5 for anything other than a high-end server architecture. It might show up in several workstations, but they've invested too much in AltiVec to make that transition. It's a lot more likely that they will use Motorola's G5 chip, possibly getting IBM to manufacture them.

    IBM has a lot more experience working with the copper wiring process as well as Silicon-on-insulator. They're able to produce significantly better yields than Motorola, which is the cause of this entire problem.

  • Megahertz sells & Joe Q. Public doesn't know what Altivec is.

    The average consumer of Mac G4s is a graphic artist who is well aware of exactly what apps and plug-ins are Altivec enabled -- and Apple knows this.
  • Well it seems lately that a lot of rumors from Apple have been true (Cube, mouse). Wait, should I be saying this?... I just got a cease and desist order from Apple. Hey, I didn't see that one coming!

    Even the samurai
    have teddy bears,
    and even the teddy bears

  • Well in my experience (your mileage may differ) the Mac OS crashes much more & harder than the Win9x OSes. Others may have different experiences, but I got damn tired of seeing that bomb on screen and having to reboot. Sure, Win stuff crashes but most of the time the app goes down first, allowing you to shutdown normally.

    I'd love to use linux all the time, but unfortunately, my addiction to Ultima Online, Diablo 2 and other games keeps me with a Win98 partion on my hard drive.

    I don't know of a ton of apps that exist on the Mac but not in Win9x. When I bought my 8500AV it was specfically because the Adobe suite didn't exist for Wintel. It was a good setup, but now with the full Adobe suite running on Windows it's much stabler (in my experience). Macs lack of memory management made it a dangerous proposition to switch between Illustrator & Photshop, usually ending in bomb. For whatever reason even quitting one program and restarting the other didn't always work because the memory allocations got all fubarred.

    Perhaps OS X and new hardware will solve Mac's reliability issues, but by then it will be too late. People like me that started with the original Macs, bought the first powerbooks (my 165c was suh-weet), and paid good money for the top end PowerPCs when they came out have simply given up on Apple. They cost alot and don't give you much more than a cool transluscent case and hardware upgrade headaches. I'll stick with Win98 and Mandrake for now on.

    The OS wars will never be won. The different sides never see each others point of view. It's kind of like me understanding why people like that candy ass, do nothing Boba Fett.

  • While I don't need anywhere near that sort of performance for most of my work, the *idea* of a quad-Altivec makes my mouth water.

    I wish Apple and IBM well!

  • AltiVec seemed like a good idea in light of Intel adding MMX support to the Pentium. But its becoming apparent that most application developers aren't terribly interested in tuning their code those special instruction sets, and what really gets the most bang for the buck, across the board, is more generalized chip updates like more MHz and bigger caches.

    Its just a case of living and learning IMO.
  • Altivec's fine, but your code needs to be specifically optimised to run on it. Hence the need for a specific Altivec-enabled Photoshop plug-in to allow the render engine to take advantage of the enhanced instruction set. It's (very roughly) equivalent to the MMX concept. A non-accellerated app runs at the same speed on an Altivec-enhanced machine as on a non-enhanced machine if it hasn't had the re-compile. Bummer!
  • "The g4 is a great processor, but i'd never consider buying a computer with *no* expansion beyond usb"

    Which is exactly why the G4, comes with 3 Firewire ports (2 external, 1 internal), 2 USB, Gigabit ethernet, SVGA port, Audio out, enough room in the case for 7 more drives beyond the standard harddrive/DVD configuration, and still have 3 empty PCI slots.

    Sounds pretty darn upgradeable to me.

  • Why'd they put a less-capable, albeit faster, processor in a HIGHER end machine?

    Marketing! Megahertz sells & Joe Q. Public doesn't know what Altivec is. All he reads is the numbers ...
  • Disclaimer; this is a rumour and we all know about Apple rumors...
    What, that you'll get sued for posting it?
  • The phrase did appear in the article, but that's not what it said. It said:

    ... Dual and Quad G4-500s in midrange machines, and G5s in high end machines ...

    Note the quad G4. G4 = PPC7400, which includes an altivec.
    #define X(x,y) x##y
  • No, Mac OS X Server is actually a dead piece of software. Mac OS X will be server and consumer in one, depending on what you install. And no, they didn't drop Multiproc. support from OS X. Why else would they have released new dual G4 PowerMacs that supplant their single proc line.
  • I am a huge mac fan, but Slashdot keeps on posting news about macs from non-reputable sites. These sites include this post, and MacOS Rumors. The Altivec chipset is great, and I find that when supported it gives great performance. I have a G4 450MP and when encoding cd's into 192Kbit stereo mp3, it chugs at about 12x completing a cd in 6mins. Apple, Motorolla and IBM need to sit down and work out the whole processor mess. They created the PowerPC processor together, keep it going!
  • I'll keep this succinct.

    A) Why can't the RISC chips keep up? I thought RISC was supposed to help INCREASE clock speeds. Is it really that Intel and AMD have that much better manufacturing technology? Or is it some other part of the architecture that's holding it back. I mean the Alpha used to be the king of clock-speed, they hit 500MHz at the .35 micron process, and I think the newest ones are still .25. However, I don't think they have 1GHz Alphas in production. What's the holdup?

    B) Why the hell would Apple ditch Altivec? It's a GOOD idea. It's not just x86 that dabbles in these instruction extensions, the Alpha has something similar to Altivec and SSE. Given the fact that 3D is going to be so big, why ditch Altivec. It's perfect for multiplying matricies, which is basically what bottles 3D at the CPU end. Not to mention all the physics stuff that uses this. Seeing as SSE boosts performance by 70% in some cases, I don't think it is a good idea to dump SSE in favor of more clock speed. Also, Altivec shouldn't really affect clock-speed, should it? I mean the Alpha got to 700+MHz with it's vector instruction set, so why not the PPC.

    C) Wouldn't an Alpha using IBM process technology be cool? Imagine a 1.5+GHz Alpha running on a .18 micron process. Given what Digital did with the .35process, would it surpise anyone if they hit 1.5GHz with a .18? However, what I want to see is an Alpha on IBM .01 micron process. Drool...
  • What's the point? The Crusoe is a low-performance low-power part. It would be quite a step down from the G4's.
  • Interesting thing about the POWER series. Apple will never be able to ship one. As I remember, the POWER2s were the size of Polaroid prints. 4096mm^2. It wouldn't fit in that cute little cube.
  • The chip IBM demoed was VERY limited. It only implemented a subset of the PPC instructions, didn't do FP, and instructions had to be "injected" into the chip via a special probe that manipulated leads on the chip. There was a joke about that. It went, "MS may make you go through hell to install software, be even THEY wouldn't go that far!"
  • What about speech recognition? Could it benefit from Altivec much? Jim
  • Disclaimer; this is a rumour and we all know about Apple rumors...

    Yeah. Expect some people to get sued.

  • Complete bull. Lots of software use Altivec. Software that needs that kind of processing power, of course. It just doesn't pay to go around optimizing your average Word's code for vector processing, you know. For video encoding/decoding, image manipulation, sound manipulation, 3D, 2D deformation, AltiVec doesn't only pay. In fact, it is the only game in town and it beats the pants of the MMXs and 3DNows of this world. But you now the single biggest piece of software that is going (or is already) to use Altivec? MacOS X. All the bells and wisthles of Quartz, its QuickTime layer (wait for QT5 to see this to the extreme), its OpenGL layer... all of them eating Altivec cycles like CRAZY, optimized for that unit and kickin ass at that. And as time passes, the more vectorization of code we will see. So see, any app running on MacOS X, like Maya, Premiere, Final Cut and whatever other software WILL be accelerated to some degree by Altivec. So please, forget about the crappy G5 rumors and rampant speculation of sites that just want to catch up hits and ads: Apple won't let VE die because it has invested quite a lot in it, it is a key part of its OS and and is at the heart of its marketing strategy (which attacks the CONTENT CREATORS as well as the consumers).
  • Building Rumors, Fooling Slashdot
    http://lowendmac.com/rumormill/2k0818.html

    You know things are wierd, when then the fake rumors start sounding
    good, if not hilarious.

    Ah yes, the iMac SEvil @ 666MHz available in Black with Red Flames
    or the Bill Gates model with limbo blue clouds motif.

  • Can't /. come up with any *real* material about Macs and Apple technologies to discuss? How about this piece [stepwise.com]for starters?

  • Motorola's G5 is supposed to be a multi-core chip. However, the design of the chip and the time that it will be available means that it *should* not be any more expensive. However, given Motorola's track record with the G4 any details of the roadmap may not be highly accurate.

    The G4e, G4+ and 7400+ are all the same thing known by different names. This is not a multi-core chip, but a processor that adds several enhancements to the G4. For example, it deepens the pipelines allowing for higher clock speeds, it adds extra AltiVec units to further increase AltiVec speed enhancements and adds an extra integer unit. The G4+ is supposed to be in test production right now, but there is no telling when it will actually be available. I think Motorola has turn to the G4+ in the hopes that it will save them from their follies with the G4. The original design had several flaws that have restricted the clock speed and even with all of Motorola's work to correct the problems, they are still having trouble.
  • Perhaps you've missed the point of what I was saying.

    "Many a true word spoken in jest".

    M.
  • chuck wasn't completely wrong. The processor is
    the PPC 7400 (more popularly known as the G4, granted)
    and the apple box bearing its name is known as the
    G4 only by virtue of the 7400's codename (which
    stuck...)
  • You Have Been Trolled
    You Have Lost
    Have A Nice Day

    BTW: you probably don't want to click the other poster's link :)
  • It sounds like clock speed is the issue. Mose web/db servers do not need the massive floating point performance something like altavec provides. If removing the altavec unit allows them to increase clock speed, integrer performance should go up linearly. Thus an altavec-less chip might allow OS X servers to compete with intel/AMD/Sun/Compaq processors

  • AMD (Score:1)

    by arielb ( 5604 )
    fortunately AMD shares tech with Alpha, IBM and Motorola so next year you'll see 2 Ghz processors, multiple cpus on one die and copper fabs 9that one is done). Plus SSE 2 for all you simd lovers
  • Expandability can be found in the high-end tower line if by expandability you meant PCI and AGP slots. Otherwise, all other Macs have USB and Firewire. Removable storage can be had in the high-end tower line. The point about having an ugly stack of peripherals laying around is conceded.
  • This is true - Mac OS X Server (which began shipping months ago, and is basically warned over OpenStep) does not support multiple processors.

    But it's really not.

    Mac OS X, which will be Apple's next gen operating system, with all the goodies like Quartz, Aqua, and the Darwin open-source foundation layer, and is probably going to go into beta within the next month or so (with a promised final version delivery by Jan 01), WILL support multiple processors. Heck, check out the Darwin web site [apple.com] to see the kernel source itself. Heck, Apple's high end machines are now all multiprocessor G4's, and even the classic Mac OS (8 and 9 at least) support a bastardized version of multiprocessor support - so surely the Mach microkerneled, BSD based Mac OS X can handle more than 1 processor.

  • Disclaimer; this is a rumour and we all know about Apple rumors...
    • Slashdot will get sued for posting it
    • Rand Race will get sued for submitting it
    • I'll get sued for posting a reply to it
    • everyone else will get sued for reading it.
    Of course, this did come from the Mac-rumor equivalent of segfault [segfault.org], so anyone who didn't post may be off the hook.

    :-)

    Every day we're standing in a wind tunnel/Facing down the future coming fast - Rush
  • ... i'd never consider buying a computer with *no* expansion beyond usb,

    So what ever happened to the two Firewire ports and the three PCI slots???

    ... plus no way to add components in the case...

    There's tons of room for more 3 1/2" media devices. All single-drive machines have at least 2 more bays open at the bottom of the machine. There's also a blank 51/4" bay at the top of the machine (if you don't have Zip) ...
  • I have secretly always wanted an Apple.
    I might have to put my dreams for a dual athlon on the back burner for a quad G5.
    But as the people at work asked me when I announced this... what would you compute on this?

    I smiled and said "Netscape."

    And after some thought... I added "ripping mp3s".

    See... a machine like this will not be wasted. :)

  • IBM doesn't do Altivec. Altivec is a great technology, but on Motorola is doing it in the G4. Altivec isn't neccesarily the most useful of technologies, and IBM sees faster clock rates selling a lot more chips to Apple.

    This rumor has some good logic behind it, truth on the other hand is arguable. Apple release the dual G4 boxes to compete in the Mhz war. It was also a cry to Motorolla to stop screwing around. The problem with the AIM alliance (Apple, IBM, Motorolla) is that they are contractually locked into non-competeing clock speeds. Meaning IBM can't release a G4 that is a ton faster than Mot'sdue to better development. But IBM can release a new version. So the G5 looks like a great chip, it's faster, and it is what Apple wants.
  • MacOS X (whenever it ships!) will have full kernel-level SMP built in so I guess Apple's customers want it, especially seeing as all Apple's high-end machines (except the Cube) are dual-processor. Dunno about MacOS X Server ....
  • Because the 128-bit registers in the G4, while powerful for certain stuff (like what you'd be doing on graphics workstations, exactly where they're talking about leaving them), don't add much to your performance in apps that don't need to do heavy duty number crunching.

    The G4 is apparently (maybe by some antiquated standard) a supercomputer. That standard measures floating point opps/sec, but most people don't need a lot of floating point performance - a more responsive Quake game comes from improved integer performance. They'd be switching from a chip optimized for floating point to one that ramps up in clock speed better, for more integer performance.

    The article claimed that they were going to leave the G4's in dual and quad processor boxes only - basically targetting them at people who are actually going to be seeing the performance boost that Altivec would give.

  • IBM has announced nothing about making any PowerPC beyond its current 750's and (licenced from Motorola) 7400's. It has said nothing about a PowerPC G5. Everything else is just rumours.

    Have a look at http://www.zdnet.com/zdnn/stories/news/0,4586,2336 967,00.html for some details of Motorola's plans for G5.
  • I believe he meant -25 C, think about the amount of heat 500MHz silicon generates when OC'd to 1000 (grab a Celeron that you don't care about sometime and try it. But have plenty of liquid nitrogen around :-)
  • Actually, Altivec handles integer math quite nicely. Altivec vectors can consist of the following:

    4 32-bit FP numbers
    4 32 bit integers
    8 16 bit integers
    8 16 bit "pixels" (see below)
    16 8 bit integers

    I believe most if not all operations are orthogonal, meaning that an operation that you can perform on one of the above data types can be expected to be able to be performed on the others as well. The pixel data type is one gears towards video processing. It's a 16 bit number containing 5 bits for each of 3 colors and one alpha bit. The difference is that overflow and underflow is handled by color. If you are adding two pixels together, and there is overflow in one or more of the color values, you don't want wraparound or bits floating over into the next field. You want the color in question to max out.

    Altivec is not really FP biased at all.
  • First I think it's strange that AIM switched to Altivec on a RISC chip because, I thought that the point of RISC was to reduce the number of instructions in order to pack more transistors onto the chip and increase performance, but then to ADD more instructions seems counter productive. Actually, that's a common misconception -- or at least a bit of idealism that is somewhat different from the actual reality of RISC processors. The idea of RISC is to reduce complexity. RISC instruction architectures often have more instructions than many CISC architectures, but the instructions are often orthogonal replications of another instruction on a different data type. The end result is a lot of simplicity in processor design. This is why even the x86 family uses a microcode engine to convert complex and no-longer supported in silicon instructions to RISC-like instructions which are the ones actually executed -- sort of an ironic throwback to the big daddy of all CISC designs, the old 2000 instruction VAX systems. Altivec is basically a dedicated DSP unit. That's why Motorola loves it. IBM however, doesn't see the need for it in high end servers, which really aren't focused on parallel number crunching. So, they don't add it to the die so that they can ramp up the clock speed. Their current weird approach to increased speed seems to be multiple CPU cores on a single die. Witness the POWER4 chip. 1 GHz clock speed with support for a 500 MHz system bus and 2 CPU cores on a single die. Too bad it's a 64-bit PPC implementation, and thus unsuitable for the 32-bit Mac OS.
  • The Transmeta Crusoe processor would not be considered for replacing the PowerPC. The Crusoe is designed for low-power emulation of x86. As the PowerPC is already a low-power device there would be no market for an low-power *emulator* of it.

    Really the only sales feature of the Transmeta products (aside from their inherent technical wizardry) is the ability to emulate other CPU's (albiet slower) in a low-power design. You won't ever see a Transmeta chip in anything other then a mobile platform.

    As to IBM making consumer PowerPC'; Yes, IBM has been manufacturing PowerPCs for Apple under license from Motorola (it's Motorola's design they're using) but not doing any design/development work on them - they're just another outsourced product in the plant along with the Transmeta chips (same plant.)

  • dumbass, finish sarcasm school before you post next time

  • Re:Slashdot =! Mac Rumor Site (Score:3)

    by KFury ( 19522 ) on Thursday August 17, 2000 @07:54AM (#848908) Homepage
    Technically, you're saying "Slashdot equals not Mac Rumor site"

    Considering that this was not actually a mac rumor, but was instead a made up story, your subject fits quite well!

    Kevin Fox

  • The G5 Roadmap (Score:3)

    by mattkime ( 8466 ) on Thursday August 17, 2000 @07:54AM (#848909)
    For info on the G3/G4/G5 roadmap look here:

    http://www.maccentral.com/news/9909/24.g5.shtml
    http://www.mot.com/SPS/PowerPC/overview/newroadm ap.pdf

    As you'll notice, NOWHERE does it say that Motorola is removing Altivec in future chips.

    The lag in processor speed is due to several factors. First, the G4 chip is not very scalable mhz-wise due to its shallow pipeline. IBM seems to have overcome this, however legal agreements are preventing IBM from producing the chip without Motorola's blessing. Future PPC designs include deeper pipelines to make scaling to higher mhz easier.

    At first, IBM was refusing to add altivec, wanting to focus on mhz instead with the ppc line. What has resulted is that IBM is simply more prepared to produce faster chips that Motorola right now.

    In all, its really a shame that this fiasco has occured. The PPC has a much cleaner design than the x86 based chips. Its a marketing miracle that Apple hasn't been held back by this. (What looks faster? Bodini blue or beige? :) )
  • Heh, More points to add to your comment.

    No marketing department in the Christian(and maybe elsewhere) world would *EVER* market a processor as 666 Mhz.

  • But the home game users are still PC people too. Maybe this will be resolved as well. Maybe Apple will try to reclaim some of that market by somehow jumping into the 3D gaming world and making an entry into serious competition. Let's see them try and win a benchmark test using Quake 3 (granted, Linux would probably have to be the testing ground due to Quake's current standings in Mac OS)

    Now IANAMU, but I am pretty sure that the gaming market for the Mac is just about dead. The only really good Mac gaming company (Bungie) sold to Microsoft to develop for the X-box instead. I would not get a Mac unless I had a specific purpose and a PC cannot do it.
  • Doubtful. Apple seems to have a differnt market focus for many things. Being in big time corporate America doesn't seem to be it. Do you know of many corporations that like having translucent colored objects?

    Apple's success in the 3D gaming market will rest largely on how ATI's 3D cards perform. That's been the bottleneck in the past. The Radeon looks good but obviously ATI and Apple are on thin ice right now.

    I've never had a problem with them in terms of hardware or compatibility. I never had to do much tinkering with the networking and worked just fine as a workstation in a heterogenous environment (Win/Mac/*ix). I never had to make them work in en environment larger than 20 computers however. I do despise their OS though. OS X looks promissing if they can speed it up a ton.
  • I bet that is one EULA agreement you would have a devil of a time getting out of. I think one has and he has a gold fiddle to show for it.

    What color do you expect the box to come in?
    Charred black?
    Red-hot?
    Melted-skin pink?

    And the tag line for the machine...

    Where would you like to go... for eterinity?

    Next thing ya know, they will have that little FreeBSD imp as a mascot.
  • Will the new 666 Apple Special Edition be available in black with flames? will the cost show up on the invoice as:
    (1) soul + 4.95 shipping?

    And everyone gives Microsoft/Intel crap for being evil?

    -----
  • Your list left out HP -- they were the original company to introduce "multimedia extension", SIMD-type instructions to their architecture.

    I forget which PA-RISC they first did this with, but they redesigned parts of the integer unit so that it could do packed arithmetic (a la MMX) at a 0% increase in die size. They literally squeezed the extra logic into the corners of the existing die frame. Pretty impressive. Sure, it only sped up a few functions (MPEG decompression was the biggie), but they essentially got it for free. Conservative, but a definite win.

    On the opposite extreme lies Alti-Vec. Alti-Vec is basically an entire vector processor sitting next to the original PowerPC core. It added 50% to the die size of the G4. 50%! It's certainly more capable than what HP or the others implemented, but Motorola paid the price. It's quite possible that they have decided the multimedia unit wasn't worth that much silicon, and have gone a different route with their next generation.

    For the curious, Intel's original MMX unit increased the die size of the classic Pentium by about 5%. They're more recent extensions no doubt add more, but I think you can see that HP and Intel took a more conservative approach to "accelerating" multimedia with specialized SIMD units.

    --Lenny

  • Rumour litmus test (Score:5)

    by mr.ska ( 208224 ) on Thursday August 17, 2000 @08:01AM (#848932) Homepage Journal
    Q: How can one accurately determine what is rumour, and what is fact, short of waiting a year or two?

    A: See if Apple sues.

  • My impression always was that the speed problem with the G4 was due to Motorola, not any inherent problems with AltiVec.

    I use a G4 at work, and AltiVec is *very* nice for certain tasks. SoundJam MP takes advantage of it, and can encode MP3s at up t 10x (haven't verified that myself). AltiVec is also used for some of the cool effects in Aqua (like maybe the "Genie" effect).

    IANAHE (I Am Not A Hardware Engineer), but I don't see why a few vector processing units would prevent a chip from reaching speeds > 500 MHz. If only IBM and Motorola would pool their technology, we might see 1GHz G5s with AltiVec soon... Maybe Jobs will kick their respective asses a bit, and it will happen...
  • Altivec can make huge diffrence for applications that it was designed for. ie, vetor based performance and large SIMD instructions (OSX's pdf thingy might take advantage of it). Things like FFTs and such should get a huge boost out of it. It's is also how the G4s have stayed 'up there' in terms of benchmarking.

    But as others have mentioned you need to have compiled for it or hand tweaked the assembly. Not a big deal. They are good chips and Altivec would appear to blow the snot out of Intels SIMD offerings.

    So yes, Altivec is a good move. It's definately a performance enhancer. Unless they keep switching the architecture every year it can't hurt them.
  • In other news, Taco353 writes: "Computer maker Trinity57 announced 2day that the new P505 will be marketed as the Q505 and feature P7 chips running at 111/222/333/444/555MHz, as well as 112/223/334/445MHz but up 2 3GHz. From 9:00 a.m. 2 3:00 p.m. you can get Q505 with P7's at 333MHz but at 3:01 p.m. Q505/P7's at 445MHz will come out. Then at 4:05 p.m. Q505/P7's at 3GHz will provide a high end (733, 1033, 10:45) machine for SE users. This will be a 24/7 sale with 300 employees, 300 computers, 300 mice, etc. Lasting until 20/20/2000 at 11:59 a.m. UDT (NIST--IP: 192.43.244.18) the sale will also have C3PO chips at 0.52KHz with 92 bytes RAM for $52.92. Special versions at 1.1/2.2/3.3KHz will also be available. Sounds good to me. What do you all think?

    --
  • While this is true, it's not the only consideration. You can also do altivec optimization at the OS level. From what I've read, many of the beautiful (and probably annoying) animations in OS X are much, much faster on a G4, probably owing to AltiVec. My guess is Apple is busy optimizing the low-level graphics routines to take advantage of AltiVec, and so all apps will get a speed boost when on a G4 Mac since much of an apps time is spent on graphics anyway.

    As for whether AltiVec was a bad decision, I'm not an expert on the subject but the impression I've gotten is more that Moto is incompetent than that AltiVec fundamentally limits clock speeds. There are rumors that IBM could produce faster G4's if they wanted to but their contracts with Moto prevents them from doing so without Moto's permission. This is touched on briefly in the article. If this rumor is true, it my represent a problem of petty bickering rather than a technological failing of AltiVec.
  • You're right, but the article mentions it as 666. Just another thing piece of info that a legitimate source would not have said.

    *sigh* I don't know what constitutes as trolling anymore. bummer. (in response to my previous post's moderation).
  • This is interesting, if true. I've seen many Mac pundits claim that Altivec was the next best thing since sliced bread. I've only had a bit of hand's on time with G4's, being an x86 or SPARC type of guy, so I'm definitely not an expert on the G4. Does anyone have any first hand experience with whether or not Altivec was a bad move?

  • OK - so /. got rooked by a parody site (note to /. editors: get the Mac fellow from Ars Technica to look over your stuff - he's excellent.)

    With that aside what ARE the alternative CPU options for Apple?

    PowerPC is doing OK as an architecture. Apple is using it in it's consumer boxes, IBM in servers, & Motorola is suplying Apple while focusing on the embedded market.

    Unfortunately there are strains on the relationship. Apple honked-off Motorola by shelving the 3rd party Mac licensing program abruptly and stiffing Motorola with a many-million dollar inventory of unusable motherboards(Apple had good reason to shelve the program ASAP but it still screwed Motorola.) On the other hand they committed to Motorola as their supplier when Motorola developed the AltiVec material. IBM politely said no-thank-you to licensing the AltiVec as it really doesn't apply to their products and would likely just get in the way of their own evolution plans for PowerPC. They've all committed to continuing the collaboration and at that level it seems to be doing well though the PowerPC "vision" seems to have succumbed to more pragmatic goals.

    The problem for Apple is that the PowerPC + AltiVec from Motorola is having trouble ramping up in speed. Apple has tried to fight the percieved speed discrepancy by pointing out that a PowerPC at some speed actually equals an x86 running at a some greater speed. All agree this is more or less true but the same as the Millenium doesn't start 'till next year no-one cares: they want comparable CPU speeds.

    Apple finally replied by shipping multiple CPU's. These work fine under MacOS 9 and will do great under MacOS X when it ships but while saying 2x500 = 1000 it still isn't really the same when it comes to bragging.

    What makes the whole thing even more ironic for all of the psuedo-knowledgable's pointing out that two CPUs aren't really twice as fast they're all ignoring there are far more fundamental issues like bus bandwidth & memory architecture that hobble Apple, at least until it's next-gen UMA2 motherboard series finally ship in a presumed few months.

    So, what are the other choices?

    Well, there have been rumors forever of Apple's MacOS-v7/8-on-x86 inhouse projects (claimed name "StarTrek".) With the move to MacOS X these now are realistic as NextStep has run on several platforms already including x86. Indeed the aborted Rhapsody strategy was actually released on x86 and the MacOS-core Darwin project is freely downloadable for x86. It wouldn't be terribly difficult for Apple to move all of MacOS X over to x86 though it would likely require abandoning all of the backwards MacOS compatability (unless something could be salvaged from the rumored "StarTrek" project.)

    Of course at that point Apple would only be selling a custom version of Unix on custom x86 boxes to the consumer market and it's doubtefull Apple could make much of a go of that in today's market.

    Next choice? Well, go to a third horse. Compaq now has the Alpha processor and it's still a speed demon and shows no signs of slowing down. There are rumors of Apple having an internal team tracking MacOS X but on Alpha. NextStep ran on Alpha so this sounds reasonable and certainly gives Apple some leverage with Motorola. The advantage of Alpha would be a screaming processor but not x86 as so to differentiate themselves.

    This would put Apple in the position of selling a custom version of Unix on custom Alpha boxes to the consumer market and this might be doable, particularly if Apple pursues a strategy of selling PowerPC or Alpha CPU's in their boxes.

    Now, there is an interesting rumor to discuss.

  • Re:devilish (Score:3)

    by PopeAlien ( 164869 ) on Thursday August 17, 2000 @08:06AM (#848957) Homepage Journal
    The fascinating thing about the 666Mhz G5 iMac SEvil, is that it actually runs infinitely faster than computers running at much higher clock speeds, but also infinitely hotter (Optimal operating temperature is 3 million Degrees F). The downside is that "Eternal Damnation" clause in the EULA, and when running Linux the Daemons can occasionally get a little out of control, spawning evil child processes, and attempting to take over the world.

    But hey, its Insanely Different(tm)..

  • G5 is not a PowerPC processor. (Score:3)

    by SoupIsGood Food ( 1179 ) on Thursday August 17, 2000 @08:07AM (#848958)
    IBM's G5 processor is not a PowerPC microprocessor. It is the silicon used to drive the S/390 mainframes: a serious hunk of electronics to complex for a single chip. It's more along the lines of a super-CISC design, optimized for speed and throughput above portability.

    IBM -does- have some sexy successors to the 750 processor (the "G3" in Apple marketing speak) and these will probably make it into future iMacs and Powerbooks. I'm more interested in their upcoming "7500+" line, which is a G4 that can be produced with a clock of 750mhz and 1ghz...we'll still have to wait until MacWorld San Francisco to see them tho.

    SoupIsGood Food
  • Does anyone have any first hand experience with whether or not Altivec was a bad move?

    Well, take a look at how much more efficient it is [apple.com] on Intel's own DSP benchmarks. Also check out the inimitable David Every's pretty good review [mackido.com] of AltiVec vs. KNI.

    Boils down to, you want to do DSP stuff, AltiVec kicks serious ass, especially if you go to the trouble to understand how to restructure your algorithms to take advantage of it. You don't, well, it's not of any real great use then.

    For Photoshop users, video compressors, sound filterers, stuff like that, it's a pretty damned big win. Errrr ... did I just describe the industries Macs dominate? Weird. You'd think Apple was actually trying to serve the needs of their core market, or something silly like that.
  • He's talking about MP3 ripping and encoding, which is accelerated by Altivec. Altivec is basically vector math processing, which allows you to perform a number of (certain types of) math operations simultaneously. Granted the main use for this is graphics operations which are CPU intensive, but there are other things (like MP3 encoding) which Altivec can also help.

    Where Altivec doesn't help is in straightforward integer operations or logic operations, such as the code behind a device driver. And of course it doesn't help with programs that haven't been compiled for Altivec...
  • Dan Knight (the operator of Low End Mac) makes no claims of truth about his "rumors". I fact he freely admits that his rumors are made up. Please if you are going to post about Mac related stuff, hire a Maccie who can filter this crap out.
  • The reason the chip would be Altivec-less is because it's from IBM. Motorolla makes the G4's with Altivec and I believe owns the rights to the technology. According to rumor however, IBM has had much better luck getting higher clock rates than Motorolla.

    Apple's in a sticky situation... they want Altivec because it speeds things up, but they need higher clock rates to compete with AMD and Intel. Motorolla hasn't been able to get high clock rates and the G4 has been stuck at 500 MHz for a year.

    It's too bad because the G4 looked so promising, but Moto has been able to deliver to all the hype.

  • LOW END MAC TROLLED SLASHDOT!!! (Score:5)

    by Golias ( 176380 ) on Thursday August 17, 2000 @07:31AM (#848976)
    Ha Ha Ha Ha Ha Ha Ha Ha Ha

    This rumor was posted on LEM by "Anne Onymus". If you bothered to check the other "rumors" from the same writer, you would have noticed that Low End Mac's "Rumor Mill" is a PARODY SITE!!!

    Dan Knight, publisher of LEM, does not take Apple rumors seriously, and under the name Anne Onymous he pokes fun at the rumor sites on a fairly regular basis.

    In other words, YHBT YHL HAND.

    In yer face, Slashdot editors! (Hehehe)

  • I have dual feelings on Altivec. First I think it's strange that AIM switched to Altivec on a RISC chip because, I thought that the point of RISC was to reduce the number of instructions in order to pack more transistors onto the chip and increase performance, but then to ADD more instructions seems counter productive.

    On the other hand my distributed net keyrate jumped when I switched one of the clients from a G3/400 to a G4/400 with altivec. The DNEC client has a portion that was written in assembly for Altivec.

    I don't know if it's a good idea or not, but unless I learn of a performance impedement I don't have a problem with it.

    LK

  • For those of you that have fogotten, IBM demonstrated a PowerPC chip running at 1 Ghz way back in 1998 (the chip ran at 1 GHz when cooled to 25 degrees Celcius, at room temperature it ran just at just under 500 MHz). Check out CNET's take on the event: IBM joins the 1,000-MHz club [cnet.com].

    The 1 Ghz PowerPC IBM demonstrated way back in 1998 was partially hand tooled. This chip broke many of the processes IBM uses to automate production of the PowerPC. Check outwhat the EE Times said about the chip at the time: IBM's 1-GHz processor taxes current EDA tools [eetimes.com].

    More recently, TechWeb states some of IBM's plans for the PowerPC: IBM Preps SOI-Based PowerPCs [techweb.com].

    To see what is available today, and what is coming in the short term future, look at IBM's product page for the PowerPC at: http://www.chips.ibm.com/products/powerp c/ [ibm.com].

    IBM intends to have out 700MHz PowerPCs for its RS/6000 line by early next year for its RS/6000 line. It would make very little sense for Apple to not start shoving these into new Macs when they become available.

    IBM has had very little trouble scaling the PPC up as it needs to for its line of servers. I really wonder why Motorola seems to be having so much trouble in the MHz race.

    OTOH, I remember 3rd party benchmarks that showed a Motorola PPC at 350 MHz smoking an Intel x86 at 500 MHz at Photoshop. And this was back in the day when x86 had MMX and PPC had no Alti-vec. FYI, the MMX instructions allowed the Intel box (running NT) to perform one or two tests slightly faster than the Apple box running Mac OS. Given this type of history, I can see Motorola being arrogant enough to think it doesn't need to keep up the MHz. But its time for Motorola to wake up and smell the coffee.

    On a related note, there is a rumor that Palm is going to drop the 68k Motorola series in favor of the StrongARM series mostly because of the MHz.

    Motorola better get some MHz action in a hurry. Despite an overall faster chip, eventually a double/triple clock speed advantage will catch up. I doubt a 1GHz T-Bird does much slower than a .5 GHz PPC, especially given Apple's slower bus.

  • Dude... if you're going to correct someone's grammar, make sure your own is correct.

    It's "you're" - short for "you are" - not "your". You wrote:

    "Hey, your in America..." which, in the context of a grammar correction message, is just sad.

  • Credibility ? Possibly.... (Score:4)

    by MrDalliard ( 130400 ) on Thursday August 17, 2000 @07:35AM (#848995) Homepage
    Whilst we know that Apple rumour sites are usually pretty lame, this does have a string of credibility to it.

    Apple have been unable to keep up in the MHz race. Whilst those of us in the know can preach that it's not a real comparison to go on about MHz, the lay-person sees these numbers and really does equate these with speed - meaning that Apple starts to lose out in the consumer market because Joe Consumer thinks that a 500Mhz G4 is slower than say, a 600Mhz Pentium (which isn't the case).

    Sales of iMacs (apparently) have started to flag, and Apple has received criticism that it isn't updating the processors in the range quick enough, which is indeed true. Some haven't been updated for about nine months. The production problems behind Motorolas G4, and the fact they can't seem to increase the speed that quickly have really been problematic - no doubt spurring on the decision by Apple to plump for IBM, who seem to have much better processor fabrication techniques.

    So - what we'll probably see is a bit of spin from an Apple show about how they've put a funky newer processor in the iMacs and that they've nearly doubled the clock speed. Those in the know again will realise that this doesn't mean the machine is running twice as quick, but the consumer will be bought over... and iMac sales will probably pick up again. Apple has to do it to stop it's flagship machine (i.e. the machine that picked it up out of the gutter) from falling off it's mast.

    M.

Slashdot Top Deals

He has not acquired a fortune; the fortune has acquired him. -- Bion

Close