GPUL on October 15 - "CONFIRMED" by Moki?

123578

Comments

  • Reply 81 of 141
    [quote]Originally posted by snoopy:

    <strong>I have been trying to understand the terminology of this industry, so I'd welcome an expert's input. I understood that sampling is where IBM makes chips in their final production facility, and samples many small runs to to see how they test out for things like yield and clock rate. It is the final step before shipping chips to be used in Apple manufacturing facility. This is what somebody told me, and if it is wrong I'd like to find out the real scoop.</strong><hr></blockquote>

    That's pretty much accurate. The goal is to find the most efficient production process. At this point, the design is final and the idea is to make sure that you can actually produce them in bulk at acceptable yield levels. This may mean small lithographic adjustments (hopefully not a major redesign) in order to increase yields. At some point in the sampling process yields will get to the point where you can begin to supply samples to external clients. Where this point is, depends on how confident you are in your production process, and in the design itself. (That is, the chips you've sampled internally meet or exceed your QA requirements and you are confident you can release some to potential customers without an embarrassment like a cache latency problem that forces you to downclock the chip.)



    [quote]<strong>So the fact that Apple has sample chips does not mean that IBM has been running final production batches of the chip. Apple would naturally have sample chips to work on just as soon as IBM has some that are representative of what they expect it to be. I would call these good prototypes, and that is what Apple has been working with, and is what the article refers to, IMO.</strong><hr></blockquote>

    Final production is just that. You've tweaked the lithography and manufacturing process and you're ramping up to full production. At that point things had better be right because you'd have to essentially shut down the line while you correct things.



    Prototypes aren't even whole chips necessarily. Prototypes can be parts of a chip that engineering wants to check to see if the CAD proves out. They can be useful in the design phase to keep from implementing a design that flat won't work. It won't tell you whether you can produce it in a cost efficient manner, however.



    [quote]<strong>Now, having said that, it is still possible that they have progress to the final production facility, and are indeed sampling right now.</strong><hr></blockquote>

    Well, that's the $64 question, isn't it?



    Moki, if you know something we don't, feel free to jump in here at any time.



    [ 10-11-2002: Message edited by: Tomb of the Unknown ]</p>
  • Reply 82 of 141
    I believe Moki has jumped in here numerous times. He has consistently said not to expect the GPUL until next fall. Personally, I believe him, more so than anyone else in this forum.



    Someone else correctly noted that IBM probably included the GPUL in this years MPF because they expect to deliver it before next years MPF. So that is when I would expect IBM to deliver the chips to Apple, sometime between July and October 2003.



    Apple will then have to ramp up production on it's machines so we should expect new GPUL machines before MWSF 2004. Probably 2-4 months before. It's my assumption that Apple has been working with IBM on the design of this chip from it's inception. This should give them a headstart on having the motherboard ready for the chip when it arrives in quantities.



    Also I expect a completely new case design with this new PowerMac. My guess would be more in line with the design of the PowerBook (grey metal - less plastic).



    These are just my guesses based on assimilating rumors from every source I can find. Wanting this to happen sooner will not make it happen. Apple knows it's customers want faster machines and would love to ship them ASAP. But wanting something doesn't make it happen before it's possible.



    Terry



    Edit: Corrected a typo and added more info.



    [ 10-11-2002: Message edited by: TBoxman ]</p>
  • Reply 83 of 141
    rhumgodrhumgod Posts: 1,289member
    [quote]Originally posted by TBoxman:

    <strong>So that is when I would expect IBM to deliver the chips to Apple, sometime between July and October 2003.



    Apple will then have to ramp up production on it's machines so we should expect new GPUL machines before MWSF 2004. Probably 2-4 months before.</strong><hr></blockquote>



    And what if these chips were not close enough to end of design last year, so they finished them this year and were discussing them at MF 2002. For all you, or anyone else here knows, these chips could be finished and ready for production. Now, wouldn't that be something.



    The only real evidence (rumors, but closest to fact we have) is the fact that Apple is testing boxes in Cupertino with Mac OS X. Would be kind of tough without a final design chip to test OS X.
  • Reply 84 of 141
    Is it a fact that Apple is testing machines with the GPUL running OS X? I have not seen mention of this fact anywhere.



    Rumor history is that everyone gets all exicted about the next big thing that when something is introduces, no matter what or when it is, rumor monger are dissappointed because it doesnt live up to the hype.



    Moki is the best source of information around here these days, and I have been on this board as a lurker for years! When he says don't expect GPUL machines until fall of next year, it means don't expect them before then. You will be dissappointed.



    Terry
  • Reply 85 of 141
    powerdocpowerdoc Posts: 8,123member
    I doubt that we will have to wait until 2004 to see this new chip. IBM is already fabbing 0,13 SOI chip like the PPC 750 fx. The project of making a SIMD unit (not the one of the Gekko) is not new, last year IBM publish a roadmap speaking of the Sahara 2 with a SIMD unit. It give them a full one year in order to develop this technology, and it's a more simple job in 2002 than it was in 2000 for Motorola. The core of the power 4 is now a mature design that have already make some evolutions, adding an independant altivec unit would not be a too big deal, perhaps more simple than adding a altivec unit to the G3 (that needed a wide internal memory data path).



    Knowing Apple i am sure they will love to present a brand new computer in January, if they can they will do. The dead end for the introduction for the GPUL is July. If Apple is waiting more time, the powermac line will be in a very big trouble.
  • Reply 86 of 141
    rhumgodrhumgod Posts: 1,289member
    TBoxman: I think I said (rumors, but closest to fact we have). I guess I wasn't clear enough.



    I am a strong believer in reading between the lines when it comes to Apple - here's a few quotes from the article <a href="http://news.com.com/2102-1001-947358.html"; target="_blank">IBM Fires up Chip Foundry</a>:



    IBM is manufacturing chips using the 130-nanometer process at the plant. The company plans to increase manufacturing quickly and reach full capacity early next year. It also plans a quick move to 90-nanometer production, which will reduce the size of the chips, allowing them to reach higher speeds and increase its manufacturing capacity yet again.



    Sounds a lot like the GPUL to me in terms of process size. Also, I find it interesting that early next year they will be reaching full capacity.



    IBM has turned the new plant into a chip foundry, which will produce a wide range of chips on a contract basis for customers in the communications or consumer electronics industries, among others.



    One of them being Apple.



    But it typically worked with only a few customers, who didn't always have access to its latest and greatest chip-manufacturing technologies.



    Can you say, Motorola?



    "We don't have the plans to have the production capacity they have (Taiwan Semiconductor Manufacturing Co.). Rather we're focusing on the high end of the market...working with the cream of the crop,"



    Apple!



    Also, I offer as proof of this not being a new technology, the IEEE Journal of Solid State Circuits, Volume 33, Number 11 from November 1998, entitled, "A 1.0-GHz Single-Issue 64-Bit

    PowerPC Integer Processor"
    which I have a copy of <a href="http://x2112.dyndns.org/personal/ieeejssc1198.pdf"; target="_blank">here</a>.
  • Reply 87 of 141
    programmerprogrammer Posts: 3,409member
    [quote]Originally posted by Rhumgod:

    <strong>Sounds a lot like the GPUL to me in terms of process size. Also, I find it interesting that early next year they will be reaching full capacity.</strong><hr></blockquote>



    I'd expect a few months between reaching full capacity and Apple shipping a machine based on it. Early 03 + a few months = 2nd half 03. Listen to Moki, he suffers the least from a lack of credibility.
  • Reply 88 of 141
    programmerprogrammer Posts: 3,409member
    [quote]Originally posted by Rhumgod:

    <strong>Also, I offer as proof of this not being a new technology, the IEEE Journal of Solid State Circuits, Volume 33, Number 11 from November 1998, entitled, "A 1.0-GHz Single-Issue 64-Bit

    PowerPC Integer Processor"
    which I have a copy of <a href="http://x2112.dyndns.org/personal/ieeejssc1198.pdf"; target="_blank">here</a>.</strong><hr></blockquote>



    I expect that that research project has little to do directly with the GPUL. It is a single-issue processor with (at least to my eye) no mention of instruction cracking or any other POWER4 goodness. The GPUL will likely be derivative from the POWER4 work done at IBM, as well as the various VMX experiments they've done.
  • Reply 89 of 141
    ptrashptrash Posts: 296member
    [quote]Originally posted by TBoxman:

    <strong>I believe Moki has jumped in here numerous times. He has consistently said not to expect the GPUL until next fall. Personally, I believe him, more so than anyone else in this forum.



    Someone else correctly noted that IBM probably included the GPUL in this years MPF because they expect to deliver it before next years MPF. So that is when I would expect IBM to deliver the chips to Apple, sometime between July and October 2003.



    Apple will then have to ramp up production on it's machines so we should expect new GPUL machines before MWSF 2004. Probably 2-4 months before. It's my assumption that Apple has been working with IBM on the design of this chip from it's inception. This should give them a headstart on having the motherboard ready for the chip when it arrives in quantities.



    Also I expect a completely new case design with this new PowerMac. My guess would be more in line with the design of the PowerBook (grey metal - less plastic).



    These are just my guesses based on assimilating rumors from every source I can find. Wanting this to happen sooner will not make it happen. Apple knows it's customers want faster machines and would love to ship them ASAP. But wanting something doesn't make it happen before it's possible.



    Terry



    Edit: Corrected a typo and added more info.



    [ 10-11-2002: Message edited by: TBoxman ]</strong><hr></blockquote>



    People are excited about this chip because 1) it's an alternative to a highly questionable future, where even the role of Motorola, Apple's main chip provider, looks highly suspect and 2) the chip apparently compares favorably to the competition, or at least improves on Apple's current chip technology. But will #2 still be the case a year, year and a half from now (I'd expect an introduction this important to be done at MacWorld SF, the trade show held on Apple's home turf, followed by shipping products a month or so later-Feb or March 2004)? And what's Apple gonna do in the meantime? (They have to make it to Winter 2004 before they can sell these new computers.)



    Oh course, both parties could be right-IBM could be fabricating the chip now, AND Apple might not put out it's new machines utilizing it until late 2003/early 2004: (From the EE Times article) "However, Apple would have to heavily rework its Mac OS, which has just gone through a major release cycle, to support 64-bit addressing. Therefore the company, which keeps a tight lid on unannounced products, might not be ready to detail its plans for the chip until the end of 2003.



    "Apple has a whole lot of work to do to fully make use of this part," said Glaskowsky of the chip.."



    [ 10-11-2002: Message edited by: Ptrash ]</p>
  • Reply 90 of 141
    eskimoeskimo Posts: 474member
    [quote]Originally posted by rickag:

    <strong>



    Motorola alleges that they have been using a 0.13µ(HiP7) process on the cores of embedded processors since April 9th 2001 and that the HiP7 process includes some parts as small as 0.07µ. This was in a press release on 4/9/01.



    My question is what the heck is taking Motorola sooooo loooonnnnnngggg to implement any of this for the G4????????



    [ 10-11-2002: Message edited by: rickag ]</strong><hr></blockquote>



    There are millions of transistors in a MPU, each transistor needs ~3 contacts sized at about .13um. It's extremely hard to fill these contacts on HiP7.
  • Reply 91 of 141
    snoopysnoopy Posts: 1,901member
    [quote]Originally posted by Ptrash:

    <strong>



    . . . (From the EE Times article) "However, Apple would have to heavily rework its Mac OS, which has just gone through a major release cycle, to support 64-bit addressing. Therefore the company, which keeps a tight lid on unannounced products, might not be ready to detail its plans for the chip until the end of 2003.



    "Apple has a whole lot of work to do to fully make use of this part," said Glaskowsky of the chip.."



    </strong><hr></blockquote>



    Personally, I don't buy into their opinion. First of all, I understand that the GPUL will run 32 bit code just fine. Second, Apple and IBM likely have been working for two years, so much OS X code written in that time period could be 64 bit ready. Third, I would imagine that BSD Unix can handle 64 bits, but I don't know that for a fact.
  • Reply 92 of 141
    ed m.ed m. Posts: 222member
    Just to get to the bottom of the quote thingie...



    Spock (Nimoy, Leonard) tells the crew, "An ancestor of mine maintained that if you eliminate the impossible, whatever remains, however improbable, must be the solution." The "ancestor" Spock quotes is Sherlock Holmes, another fictional character well-versed in logic. Leonard Nimoy and co-star Plummer, Christopher have both played Holmes on stage and screen. Also, director Meyer, Nicholas is the author of several Sherlock Holmes novels, including "The Seven Per-cent Solution," considered by many to be the best Sherlock Holmes story not written by Sir Arthur Conan Doyle.\t



    Taken from here:



    <a href="http://www.amazon.com/exec/obidos/tg/stores/detail/-/dvd/B00005OAZY/quotes-trivia/ref=pm_dp_ln_d_5/103-0367551-7114248&quot; target="_blank">http://www.amazon.com/exec/obidos/tg/stores/detail/-/dvd/B00005OAZY/quotes-trivia/ref=pm_dp_ln_d_5/103-0367551-7114248&lt;/a&gt;



    --

    Ed
  • Reply 93 of 141
    I still come back to this question:



    Why did Apple release such a heavily reworked PowerMac line when it did when there really wasn't anything about the processor or logic board that necessitated this major change in cooling, etc.?



    Still, the only good answer that I can conjure up is that they wanted to give the design the ultimate shakedown in the field before they stick in the stuff that really needs the new design. That shakedown might need a few months but in no way does it need a year.



    Couple this with the statement that new Macs won't boot OS 9 come 2003 and I think it adds up to a new processor (presumably from IBM) sooner as opposed to later.



    Lastly, is there really any reason to think that the new chip actually needs a 64 bit ready OS to function? Couldn't 64-bit come later and maybe only for servers or a new workstation line? My two cents.
  • Reply 94 of 141
    ed m.ed m. Posts: 222member
    [[[Personally, I don't buy into their opinion. First of all, I understand that the GPUL will run 32 bit code just fine. Second, Apple and IBM likely have been working for two years, so much OS X code written in that time period could be 64 bit ready. Third, I would imagine that BSD Unix can handle 64 bits, but I don't know that for a fact.]]]





    Good point Snoop...



    The simple fact is that *if* IBM had any intentions of attracting Apple as a possible customer, it's highly likely that IBM would have been working closely with Apple to insure that the chip could at least be an option for Apple. What's really odd is that *my* sources have been unusually quiet lately. Still, it's been echoed from various sources -- No one expected Apple to still be using the current G4 part in 2002. They were supposed to be on another CPU by now. Rumor also has it that this gp-ul (originally meant to be typed in lower case btw) was in the works for quite some time; at least 2 years from what I'm told. AND they were working WITH Apple. On another note, It's encouraging that IBM actually mentioned Apple as a possible customer for the chip. I remember a quote that was posted to the Web a while ago, I'm not sure how long, but from what I remember, it apparently was a quote from Steve Jobs talking about system performance with respect to the competition. It mentioned something to the effect of "The G5 will be Apple's savior." I know it sounds kinda corny and maybe even made up, but like I said I read it a long time ago on some obscure message board. Does anyone else remember reading that or am I the only one? Perhaps someone can even determine the origin of it.



    --

    Ed
  • Reply 95 of 141
    snoopysnoopy Posts: 1,901member
    [quote]Originally posted by Ed M.:

    <strong>Just to get to the bottom of the quote thingie...



    Spock (Nimoy, Leonard) tells the crew, "An ancestor of mine maintained that

    if you eliminate the impossible, whatever remains, however improbable,

    must be the solution." The "ancestor" Spock quotes

    is Sherlock Holmes . . .



    </strong><hr></blockquote>



    Also being a Star Trek fan, I'm surprised I did not know about Spock's

    quote of Holmes. The two do think alike in many ways.



    [ 10-11-2002: Message edited by: snoopy ]</p>
  • Reply 96 of 141
    mokimoki Posts: 551member
    [quote]Originally posted by TBoxman:

    <strong>I believe Moki has jumped in here numerous times. He has consistently said not to expect the GPUL until next fall. Personally, I believe him, more so than anyone else in this forum.</strong><hr></blockquote>



    Hmmm, I don't think I said next fall... there are a number of variables here, to to be bluntly honest, no one knows for sure when these suckers are going to ship in Macs, including the people at Apple who are working on them.



    They have target dates, sure -- but target dates frequently slip -- so predicting when they will be released as something you can buy is quite a tough thing to do.



    It is easier to say that they will *not* ship at MacWorld/SF, and that they will ship before 2003 ends. Further than that, other than optimistic projected dates from project managers, no one knows when it'll all come together.



    Just because IBM ships the chips doesn't mean that a hardware/software platform will instantly spring up around 'em.
  • Reply 97 of 141
    Not wanting to throw Dorsal posts into the mix, but he did state several months ago that Apple was shipping test mules in RS/6000 form factors booting Darwin only. Maybe the GP-UL is what he was hearing about or POWER4 based mules to get Darwin to work on. I checked IBM's site and they ship workstations with POWER3 & POWER4 chips. If Darwin is booting on it, sounds like a really good start.



    I still think it will be around MWNY03 (or is it Boston).
  • Reply 98 of 141
    ed m.ed m. Posts: 222member
    [[[I still come back to this question:Why did Apple release such a heavily reworked PowerMac line when it did when there really wasn't anything about the processor or logic board that necessitated this major change in cooling, etc.? ]]]



    Agreed. I thought the same thing when I first saw the innards.





    [[[Couple this with the statement that new Macs won't boot OS 9 come 2003 and I think it adds up to a new processor (presumably from IBM) sooner as opposed to later. ]]]



    Maybe.



    [[[Lastly, is there really any reason to think that the new chip actually needs a 64 bit ready OS to function? Couldn't 64-bit come later and maybe only for servers or a new workstation line? My two cents. ]]]



    It doesn't need a 64-bit OS. What is gives is time... time for Apple and developers to shift to 64-bit when it's necessary to do so. However, I like to remain optimistic.. That said, I wonder if Apple has been maintaining a 64-bit OS in parallel with the 32-bit version.



    --

    Ed
  • Reply 99 of 141
    davegeedavegee Posts: 2,765member
    [quote]Originally posted by moki:

    <strong>



    Hmmm, I don't think I said next fall... there are a number of variables here, to to be bluntly honest, no one knows for sure when these suckers are going to ship in Macs, including the people at Apple who are working on them.



    They have target dates, sure -- but target dates frequently slip -- so predicting when they will be released as something you can buy is quite a tough thing to do.



    It is easier to say that they will *not* ship at MacWorld/SF, and that they will ship before 2003 ends. Further than that, other than optimistic projected dates from project managers, no one knows when it'll all come together.



    Just because IBM ships the chips doesn't mean that a hardware/software platform will instantly spring up around 'em.</strong><hr></blockquote>





    "target dates frequently slip"



    Sure... you expect us to buy that?!?!



    Hey Moki, why don't you just peddle your logic somewhere else... We ain't buying it!



    Fact is as much as everyone here hates to admit it... Moki is dead on. Nobody but the top brass at Apple and maybe the hardware folks working on the systems know when this CPU will turn up at the Apple Store near you. Heck this CPU might not even be 'the next big thing' maybe it's gonna be 'the next next big thing'. My faith in MOT is all but gone so I do hope that gpul will be soon but that's just it... it's HOPE.



    I can tell you one thing... I've known of the gpul for well over a year now and it wasn't till Moki made mention of it a few months ago till I started to get REALLY excited (and then the MPF news just pushed me over the top). Yea I was excited when I 1st got wind of it but I also knew that:



    A - Most people wouldn't believe me and I couldn't /wouldn't be able to prove it

    B - I really didn't have too much to tell

    C - At the time I just didn't feel 'right' talking about it.



    Every once and again I'd search out gpul and turn up a few spanish web sites (don't know what it means) but I knew they weren't talking about an IBM PPC processor.



    If I've known about it for a year I gotta think it's been in the works for quite a while before that... So when people say 'two years in the works' I don't doubt it for a moment.



    My point is these things just don't happen over night and no even 'over year'... Knowing about the gpul for as long as I have has given me great insight on just how long this stuff can take.



    So when you read some silly rumorz site report that 'Apple is so mad at XXXX they are going to go to YYYY for the next CPU' or Apple is gonna play 'XXX' against 'YYY' because (whatever) well it ain't true... Or if it is true it's gonna be QUITE SOME TIME before you or I see the impact of it.



    Dave



    [ 10-11-2002: Message edited by: DaveGee ]</p>
  • Reply 100 of 141
    mokimoki Posts: 551member
    [quote]Originally posted by DaveGee:

    <strong>Fact is as much as everyone here hates to admit it... Moki is dead on. Nobody but the top brass at Apple and maybe the hardware folks working on the systems know when this CPU will turn up at the Apple Store near you. </strong><hr></blockquote>



    I don't think the top brass know at all -- they know when they *want* it to come out, but let's face it, it can't ship until it is done. Something as complex and interdependent as a computer system can't be accurately predicted down to the month.



    They have target shipping dates for hardware and software, but if you think these dates are generally reached, well, I have a bridge to sell you...



    I expect you'll see GP-UL based Macs shipping next summer or fall... it is possible that it will ship before that window, but I find it very unlikely. If it slips too far past fall, well, Apple is in deep shit.
Sign In or Register to comment.