Intel shows new chips, outlines platform directions

1234579

Comments

  • Reply 121 of 177
    snoopysnoopy Posts: 1,901member
    Getting back to the RAM requirement for running 32 bit application on a 64 bit system. It suddenly hit me what I learned in CS kindergarten. The smallest addressable chunk of RAM is a byte, 8 bits. So there should be no problem putting integers into 32 bit chunks of RAM. Of course the address pointers will be 64 bits, but we don't need so many of those. Now I see what some of you have been saying.
  • Reply 122 of 177
    melgrossmelgross Posts: 32,977member
    Quote:

    Originally posted by TenoBell

    But HP nor Dell are nearly as secretive as Apple. And both are far more predictable.







    No I was saying if we go with current appearances Macintel software is only being translated for 32 bit.



    The other part of the story is that Intel is really lukewarm on EM64T. They've been forced to support it because Microsoft adopted it for XP-64. An Intel spokesman even went as far as to call EM64T a feature extension of x86, and not much of an architecture advancement.



    Intel really wants to push IA-64.







    The problem is that the baggage comes along with the chip. From what I've been reading x86 has major design inefficiencies that are none existent in newer architectures.







    Just pointing out that developers are transferring code on a chip that won't even be in use next year. That is significant in that Intel will be using a different architecture.



    Yes other PC companies will use Intel's newest chips, but they will be far more obvious with what they will use.



    My point is that Apple has not been obvious at all. Everyone assumes it will be obvious, but Apple has not confirmed that.









    You are right that Intel will be forced to continue with x86 because the majority of the PC industry won't leave it.



    But Intel is developing other architectures besides x86.



    Apple has little stake in x86 and at this point and has options.







    Our x86 line cannot compete with Power.







    It sounds as though you think I am with the people arguing Apple should stick with PPC. I am not.







    There are still other options within Intel.




    Well, Apple wasn't as unpredictable when it has a processor line that could be relied upon. You could predict what Apple would do.



    January they would have a new machine(s) and chip.



    March/April would be a speed bump, perhaps a larger Hd, possibly more memory.



    July would be more new machines, major chip and parts upgrades.



    September would repeat March/April.



    Apple is only doing 32 bit on x86 now because the tools for 64 bit apparently haven't been available as they will be soon. hopefully by the time the first Macs that need it come out, it will be settled. Most likely with Leopard.



    64 bits are an extension to the x86 architecture. That doesn't make them elegant, as in the G5, but it seems to be effective.



    Intel is committed to this. They would love to see x86 disappear altogether. That was their original idea. But it isn't going to happen. It's possible that the Itanium might disappear instead.



    The baggage isn't that important. Apple doesn't have legacy software or hardware that uses it. They will just ignore what isn't relevant.



    The chip isn't important to using the code. All PC companies will be going through the same transition to the new chips. This is, for the most part, transparent. Some changes always have to be made for efficiency and such, as well as some compatibility.



    We have gone through those same problems every time Apple went to a newer PPC.



    I'm not so sure how much more obvious they will be. The same chips will be available to everyone, and Apple will have the same choices for the same class of machines. Apple will be constrained to the same extent that anyone else will be.



    The only question here will be whether Apple will be on the leading edge or somewhere in the middle. Either strategy has its own risks and benefits.



    What hi-power architecture is Intel working on other than x86 and IA-64? The other architectures are for handheld use. They don't compete in this space.



    You took your own quote out of context. He was comparing the Power to the Itanium, and explaining why they would continue producing it. That was why he mentioned the 32 bit line. The 32 bit line is now becoming a 64 bit line.



    The quote:



    "Long term, the architecture Itanium needs to aim at is [IBM's] Power line. We have nothing in our existing 32-bit line capability that can compete with Power. It's a very high performance line." CEO Paul Otellini



    I'm not saying that you are saying that Apple should stick with the PPC. I'm commenting on your other statements.
  • Reply 123 of 177
    melgrossmelgross Posts: 32,977member
    Quote:

    Originally posted by THT

    Architectures don't have potential for growth, only business models do.



    Yes, exactly.
  • Reply 124 of 177
    thttht Posts: 3,952member
    Quote:

    Originally posted by TenoBell

    That is actually the crux of the problem too many transitors. It becomes an art of diminishing returns.



    Umm... you are going to be using a billion transistor CPU and you'll like it!



    Seriously, in terms of performance, too many transistors is not the problem, it is what gives it more performance. This will be true for the next several years.



    The only time diminishing returns comes into play is in the business model, not in the performance. A billion transistor processor at 65 nm would have about 400 to 500 sq mm of die area, reducing fab capacity 4 times over Yonah, making it much more expensive to make per fab. And indeed, Intel will charge 3 or 4 thousands dollars for a Whitfield CPU. Next node at 45 nm though, you can buy one for $500.
  • Reply 125 of 177
    tenobelltenobell Posts: 7,014member
    Hmm, interesting.



    I had no idea others would so vehemently support Apple using standard Intel x86.



    Especially at a time when Intel is prepared to move from x86 and with its considerable weight support a new superior CPU architecture and ISA.



    Apple has no reason to lock itself into x86. Apple is in a good position to be the beneficiary of Intel's new architecture and ISA.



    We should support Apple turning the Macintosh into a general PC box? Interesting indeed.



    Quote:

    Apple is only doing 32 bit on x86 now because the tools for 64 bit apparently haven't been available as they will be soon. hopefully by the time the first Macs that need it come out, it will be settled. Most likely with Leopard.



    I wasn't trying to say why that was, I was just saying it was. Also within the context of Intel's lukewarm and forced support of EM64T.



    Quote:

    We have gone through those same problems every time Apple went to a newer PPC.



    Not exactly the same. x86 is all about supporting legacy, there will be useless and unnecessary transistors on the chip.



    Quote:

    The same chips will be available to everyone, and Apple will have the same choices for the same class of machines. Apple will be constrained to the same extent that anyone else will be.



    This does not follow Apple or Steve Jobs history and I don't understand what benefit it will be now.



    The Macintosh would be the same machine as a Dell except for the OS. What benefit is that?



    Quote:

    You took your own quote out of context. He was comparing the Power to the Itanium, and explaining why they would continue producing it. That was why he mentioned the 32 bit line. The 32 bit line is now becoming a 64 bit line.



    They will continue producing the Itanium because the x86 line is about to go 64 bit? That's not what I got out of that statement, can you clarify?



    Apple is going to do what ever it is going to do, irregrdless of what we say here.



    We just have to wait.
  • Reply 126 of 177
    tenobelltenobell Posts: 7,014member
    Quote:

    Architectures don't have potential for growth, only business models do.

    Yes, exactly.



    Growth has more than one meaning or context.



    Synonyms: development, progress, advancement.



    IBM?s Power 5 architecture is a _____ of its Power 4 architecture.



    A. growth

    B. development

    C. progression

    D. advancement
  • Reply 127 of 177
    tenobelltenobell Posts: 7,014member
    Quote:

    Seriously, in terms of performance, too many transistors is not the problem, it is what gives it more performance. This will be true for the next several years.



    Certainly I"m no expert on microprocessors.



    From my understanding this is the very reason the P4 hit its wall. Too many transistors, too many pipe lines, too much heat, too much power.



    The reason IBM is able to advance Power and the Cell more specifically is because of the reverse engineering of the transistor model. Only using transistor needed for maximum performance. They have short pipelines, in order execution, and no branch prediction. These chips are rated as high as 3.6 Ghz and consume less than 5 watts.



    From what I understand these strides are very difficult to make with current x86 architectures.



    From what I've read Intel is able to hype Pentium M because of its size in the market place. But in reality other architectures will surely beat Pentium M in performance and power consumption.
  • Reply 128 of 177
    sunilramansunilraman Posts: 8,133member
    Quote:

    Originally posted by TenoBell

    Certainly I"m no expert on microprocessors.



    .............



    From what I've read Intel is able to hype Pentium M because of its size in the market place. But in reality other architectures will surely beat Pentium M in performance and power consumption.






    Morning all, I'll jump in here. The Pentium M is gaining momentum and 'reputation' not just because of its size in the marketplace, there is a reason why it got to that size as well



    we're not saying oh Intel is the bomb now. I am using an AMD gaming rig right now, and it's sweet.



    what we are trying to do is look at why apple Switched. basically, IBM and Freescale have not met their business requirements. Steve & Co are now hoping that Intel will meet their business requirements in the next 5 years so that apple can produce highly profitable digital computers and devices to make the most of the iPod momentum.



    which chip is better at what is a moot point and computer science general knowledge, while its fun learning that, really does not always put us in a better position to know what is going to happen.



    i'll say it again, my view is that the key now for apple is their operating system and COCOA APIs. and how good and useable xCode 2.1++ is for developers.



    based on what we have learnt, the idea is now that Apple through third party collaborations prepares the hardware. Apple now puts on the Mac os X which through cocoa and frameworks such as Accelerate.framework abstracts out all the lower level stuff so that Joe/Jane Smart Developer can focus on xCode and Cocoa and the big/little endian issues.



    as mac enthusiasts this is where our interest lies....



    1. hardware: chips that run fast, use the latest technology, are reliable, and generate less heat and waste less energy



    2. operating system: a sophisticated enough mac os X that can fly on everything from dual-dual-g5s or 4-core-x86's all the way 'down' to a 32bit powerpc g4... this is predominantly apple's responsibily, in terms of the OS and Cocoa APIs



    3. development environment: a robust, lively, engaging, exciting developer community that delivers great apps that enhance the 'reference' iApps that Apple puts out, by making the most of Cocoa APIs and writing clean, abstracted, *hardware agnostic* code. and giving appropriate feedback to apple on how to continuously improve those APIs





    PPC and x86 will have to live together for a while at least until the end of 2007 and likely longer too



    While Cell and Power5, Power6, etc is great for super-performance by optimising your low level code for specific purposes eg. high-end gaming or supercomputer servers, etc, etc, Apple's business focus is moving away from "specialised computing" to a portfolio of fun, exciting, easy-to-use Mac experiences that can be *easily enhanced* by talented 3rd party Mac developers. And here i am definitely hinting at a device somewhere between a sub-iBook and the iPod which is definitely a possibility in the next 3 years, particularly with some of the stuff Intel showed
  • Reply 129 of 177
    thttht Posts: 3,952member
    Quote:

    Originally posted by TenoBell

    From my understanding this is the very reason the P4 hit its wall. Too many transistors, too many pipe lines, too much heat, too much power.



    It wasn't too many transistors. It was poor design. Moreover, it is the Prescott implementation of the Netburst architecture that bombed. Its execution pipeline was extended in hopes of getting to 5 GHz, a rather extreme strategy and it bit Intel in the ass.



    They are now being sensible about it with Merom. Ie, the engineers seem to be running the Intel house now rather than the marketers.



    Quote:

    The reason IBM is able to advance Power and the Cell more specifically is because of the reverse engineering of the transistor model. Only using transistor needed for maximum performance. They have short pipelines, in order execution, and no branch prediction. These chips are rated as high as 3.6 Ghz and consume less than 5 watts.



    You're in error. Cell and Xenon burn 50 to 80 Watts, have 21 stage pipelines, and do have branch prediction, but likely poor. And they didn't reverse any transistor model or anything. What they did was a design tradeoff.



    What they did was sacrifice extremely important performance features (OOOE and wider-issue) for clock rate. That comes with a price: for most general purpose code, a 1.6 GHz PM or G4 is as fast as a 3.2 GHz Cell. They also put in a bit of custom design in the circuitry including dynamic logic circuits. That will come with a cost as well in longer development times for improvements on the chip.



    Yonah and Merom will be tons better for Apple.



    Quote:

    From what I understand these strides are very difficult to make with current x86 architectures.



    Nope. The thing that was done for Cell can be done for x86. Heck, Intel did it before IBM, and did it with more state-of-the-art features. Cell is like a Northwood P4 with all of the features needed for good single threaded performance taken out.



    Quote:

    From what I've read Intel is able to hype Pentium M because of its size in the market place. But in reality other architectures will surely beat Pentium M in performance and power consumption.



    Nope. Pentium M is the performance/power king in the PC market.
  • Reply 130 of 177
    thttht Posts: 3,952member
    Quote:

    Originally posted by TenoBell

    Growth has more than one meaning or context.



    Yes, but it is the only one that matters. More specifically, that matters to Apple.



    Quote:

    IBM?s Power 5 architecture is a _____ of its Power 4 architecture.



    A. growth

    B. development

    C. progression

    D. advancement




    Apple can sell a profitable Powerbook for $1500 with a ___.



    A. Power 5

    B. Yonah

    C. Cell

    D. G5



    It's really simple. If IBM can't produce an affordable CPU with appropriate performance per watt for Apple, all the potential of the ISA does not matter anymore. On the other hand, Intel's very life is dependent on them doing so.
  • Reply 131 of 177
    melgrossmelgross Posts: 32,977member
    Well, well, well!



    To put a fly in the ointment, as they say, I recieved this page in an e-mail from my Forbes account. It's under shares to dominate in trading Monday. When I clicked on the link, this is what showed up.



    Interesting dates, would you say?



    http://www.forbes.com/2005/08/26/app...eonstocks.html



    Now that I went to 'insiders front page I see they have some info about it there as well.



    I would have caught that if I wasn't struggling all night long to install 10.4 on this very machine, the last holdout here.
  • Reply 132 of 177
    tenobelltenobell Posts: 7,014member
    Quote:

    Nope. Pentium M is the performance/power king in the PC market.



    Well....ok, cool.



    That's not what I've been reading from other sources.



    We shall see.



    Quote:

    Apple can sell a profitable Powerbook for $1500 with a ___.

    A. Power 5

    B. Yonah

    C. Cell

    D. G5



    My point wasn't about Power at all. No I'm not arguing for Apple to stay with Power.



    The point was the term growth can be used in different context.



    Quote:

    Freescale Semi to supply Apple Computer with PowerPC processors through 2008



    It could be a back door contingency incase the Intel transition doesn't work out. That may be a secondary reason.



    But more than likely the primary reason is so Apple can have replacement parts for the millions of G4 that will run for years.
  • Reply 133 of 177
    telomartelomar Posts: 1,804member
    Quote:

    Originally posted by TenoBell

    That's not what I've been reading from other sources.



    Your sources are bad, read something credible. The pentium M easily has the highest performance/watt compared to POWER5, Itanium, current Athlons and Opterons, PPC970 and Pentium 4's and their dual core variants.



    Links have been posted to benchmarks attesting to this previously. It really is a gorgeous effort of processor design and if Merom really does only draw 5 watts (I know 30 was being bandied about 6 months ago so this surprises me) it will make the architecture even nicer.
  • Reply 134 of 177
    tenobelltenobell Posts: 7,014member
    Quote:

    Your sources are bad, read something credible. The pentium M easily has the highest performance/watt compared to POWER5, Itanium, current Athlons and Opterons, PPC970 and Pentium 4's and their dual core variants.



    Well??.



    I know of others who would just as passionately claim that you are wrong.



    Some would claim that AMD has much better performance than Pentium M. They would cite tests, chart/graphs, and benchmarks. Have a whole speech about data I/O and bus speeds.



    Really it seems to come down to opinion and dare I say preference.



    Quote:

    my view is that the key now for apple is their operating system and COCOA APIs. and how good and useable xCode 2.1++ is for developers.



    Largely and for the most part I agree with this. With all things being equal it should come down to who has the better software.



    At the same time one of the benefits of using the Mac platform is that Apple is not in total lock step with the larger PC industry.



    To a degree the Mac platform has its own independence and Apple is able to exercise that independence when it sees fit. Whether this is with software or hardware independence has its disadvantages as well as its advantages.



    The ability to adapt its hardware gives its software new abilities. Apple has been able to do this in ways that the larger PC OEM?s have not been able.



    I am wary of Apple locking itself in with everyone else, and having to follow what everyone else does. Even in the hardware that will effect what they are able to do with software.



    With this transition there are other alternatives besides Power or x86 that Apple could take full advantage of. Alternatives that would allow Apple to work directly with Intel and still remain out of lock step with the industry at large, giving Apple room to innovate, hardware as well as software
  • Reply 135 of 177
    Except for two good reasons: cost and power consumption. No other chips are produced at the volume of x86 processors that offer near the performance. And while the G5 holds its own in terms of IPC, it's not gonna be stuffed in a Powerbook any time soon.



    the alternatives you speak are not really alternatives. And Apple being out of lock step with the PC industry is actually what prompted the switch to x86. Not just the x86 of today, mind you, but the x86 of the next few years ahead as well.



    As THT pointed out:



    Apple can sell a profitable Powerbook for $1500 with a ___.



    A. Power 5

    B. Yonah

    C. Cell

    D. G5



    What alternative do you speak of? Sparc? Xscale?
  • Reply 136 of 177
    telomartelomar Posts: 1,804member
    Quote:

    Originally posted by TenoBell

    Some would claim that AMD has much better performance than Pentium M. They would cite tests, chart/graphs, and benchmarks. Have a whole speech about data I/O and bus speeds.



    Really it seems to come down to opinion and dare I say preference.




    Problem is we are talking perf/watt and AMDs offerings are aren't as efficient by comparison for the equivalent performance. Don't get me wrong compared to Netburst they were considerably better but AMD is in very serious trouble where Pentium M is concerned and where Intel's roadmaps are headed.



    Until 2007 AMD is going to find they rely on their bandwidth argument.
  • Reply 137 of 177
    sunilramansunilraman Posts: 8,133member
    Quote:

    Originally posted by Telomar

    Problem is we are talking perf/watt and AMDs offerings are aren't as efficient by comparison for the equivalent performance. Don't get me wrong compared to Netburst they were considerably better but AMD is in very serious trouble where Pentium M is concerned and where Intel's roadmaps are headed.



    Until 2007 AMD is going to find they rely on their bandwidth argument.




    I'm sure AMD is hard at it, luckily now in the "desktop performance" and "gaming rig" market they are holding their own, at least my impression of it.



    But yeah, they need something to match the Pentium M by end of 2006.



    Currently Pentium Ms are expensive if you want to set up your quiet desktop PC. Laptops though yes, pentium Ms doing well, although honestly i am very curious about athlon mobile and turions.... what is happening there and how they fare....
  • Reply 138 of 177
    brendonbrendon Posts: 642member
    Quote:

    Originally posted by TenoBell

    Hmm, interesting.



    I had no idea others would so vehemently support Apple using standard Intel x86.



    Especially at a time when Intel is prepared to move from x86 and with its considerable weight support a new superior CPU architecture and ISA.



    Apple has no reason to lock itself into x86. Apple is in a good position to be the beneficiary of Intel's new architecture and ISA.



    We should support Apple turning the Macintosh into a general PC box? Interesting indeed.



    I wasn't trying to say why that was, I was just saying it was. Also within the context of Intel's lukewarm and forced support of EM64T.



    Not exactly the same. x86 is all about supporting legacy, there will be useless and unnecessary transistors on the chip.



    This does not follow Apple or Steve Jobs history and I don't understand what benefit it will be now.



    The Macintosh would be the same machine as a Dell except for the OS. What benefit is that?



    They will continue producing the Itanium because the x86 line is about to go 64 bit? That's not what I got out of that statement, can you clarify?



    Apple is going to do what ever it is going to do, irregrdless of what we say here.



    We just have to wait.




    The Macintosh would be the same machine as a Dell except for the OS. What benefit is that?



    "It's all about the software" SJ. OSX is light years ahead of Windows, and only getting better. Security is a huge advantage for OSX and that is not from the hardware. About the only weakness that I can see is Applescript that could come more of age, but I'm thinking about that becoming the equivalent of VB for the Mac, what would be better is to have a VB for the Mac. A light weight programming language, that makes it easy to write programs for the Mac. Not allot of folks do C or Objective-C. So you see I talk about software, hardware will be fine. On the PPC it was just about Apple that was adding optimized code to GCC for PPC, now flip to X86 and hundreds of folks are adding optimized code, even that impacts Free BSD. So if you are looking at the cost of using hardware that no one else is using I would have to say add that to the list of costs.
  • Reply 139 of 177
    melgrossmelgross Posts: 32,977member
    Another point must be made. What we see now isn't what is going to be available mid 2006, and not even close to what will be out in 2007.



    While the arguments are true that it's a close call now, it won't be by the time Apple moves over. Even the Intel/AMD differences will change substantionaly by then.



    Intel simply has more resources than AMD, even with IBM's help, could ever muster.



    At some point AMD will lose its lead. It's inevitable. I'm sure they're concerned about as well, though they, of course, can't admit it. Ergo, the lawsuits. They're hoping it will distract Intel enough to make them lose their focus.
  • Reply 140 of 177
    jousterjouster Posts: 460member
    Quote:

    Originally posted by TenoBell

    ...Apple has no reason to lock itself into x86. Apple is in a good position to be the beneficiary of Intel's new architecture and ISA....



    Isn't the whole point to avoid being tied to another low-uptake architecture that doesn't generate enough money/sales to power future improvements?



    Quote:

    Originally posted by TenoBell

    We should support Apple turning the Macintosh into a general PC box? Interesting indeed.



    Sure, from a hardware point of view, and for the reason I gave above. The OS is enough of a differentiation from Dell et al.
Sign In or Register to comment.