Power5 based 975 90nm chips to ramp in May?

135678

Comments

  • Reply 41 of 148
    Quote:

    Originally posted by Mr. MacPhisto

    I haven't heard, but I expect the 90nm Power5 derivatives (they are not badged officially yet) to have 1MB of L2 cache. I fully expect them to be in the next PowerMac revision - and I expect that announcement at WWDC.



    Glad to hear there may be some actual corroboration of my "straight to 3GHz with the 975" theory.



    Mr. MacPhisto - any news on PCIe implementation?
  • Reply 42 of 148
    Quote:

    Originally posted by Programmer

    Lets get 90nm going before we start talking about 65nm. I know its coming, and I know what they're planning for it, but first they have to deliver on their current plans.



    I hope Freescale has gotten its act together. Dual core doesn't interest me, but an on-chip memory controller and HT or RIO I/O bus with a fast (i.e. 2 GHz) G4 core would be a very compelling PowerBook chip.




    I agree about the G4, especially if the L2 cache is bumped up to 1MB. To be honest, a G4 at 2GHZ with fast bus support and true DDR implementation coupled with 1MB of L2 cache would be a killer chip in the PowerBooks and may force the G5 to stay in the desktop lineup with the G4 becoming Apple's portable chip, at least for a time. I'm not sure if Apple would be interested in the dual-core chips or not. Obviously not for the PowerBooks (not now, anyways), but for some other application.



    As for the 65nm process at IBM. They seem to feel that the transition will be easier than the 90nm transition has been. The hope is that 65nm chips could be fabbed by the end of the year. Fishkill is ready to go @ 65nm and much of the technology has been implemented at 90nm. However, the dual-core chips would not be part of the initial fab at 65nm.



    As for the multi-threading, that should be featured in the 90nm version of the Power5 derivative. I've been told there will be chips capable of being placed in notebooks - but don't know if Apple will use them or not.
  • Reply 43 of 148
    Quote:

    Originally posted by Ensign Pulver

    Glad to hear there may be some actual corroboration of my "straight to 3GHz with the 975" theory.



    Mr. MacPhisto - any news on PCIe implementation?




    I'll admit, I'm not a chip guru - but I'd think that decision would like with Apple. I only know people at IBM and information I get is usually fairly generalized when concerning their work. For instance, I know that the Power5 derivative will go beyond 3GHZ upon introduction, but I don't know how far beyond. I've known since last year that it was scheduled to be produced this summer. I didn't know, at first, whether it would be the chip to take Apple past 3GHZ. There was some thought amongst my friends that the 970FX could be taken up that high - although it would require a good cooling system, but that was based on the 90nm transition not hitting the snags it has hit.



    So, I know nothing about PCIe implementation, although I hope it will come with the next revision. It is my guess that revision will come at WWDC - just because of the one year deadline Jobs placed on Apple.



    From what I've pieced together, however, it is my belief that Jobs gave that date as a conservative estimate, hoping that IBM would deliver 90nm 970FXs at 3GHZ for introduction in January and shipment sometime in February. However, yields were low and there was no way the 970FX could get to 3GHZ when they were having trouble jacking it beyond 2GHZ. As I've said, my feeling is that the 2.5GHZ chip was meant for the xServes. I think they figured out that 3GHZ wasn't a possibility sometime in the fall, but still hoped for 2.5GHZ on a PowerMac update - but even that wasn't happening when they had to shift the xServe to 2GHZ and forgo PowerMac updates. There wouldn't have been enough chips to move the PowerMacs to 90nm and there weren't substantial benefits (for instance, L2 cache remained the same).



    So, in a nutshell, Steve knew that the Power5 derivative would be ready in a year and that the 90nm process should have its kinks worked out by the summer (and things are coming together - no seriosu problems are expected with that fab). That's my guess about Steve, based on the information I have from IBMers.



    Of course, the possibility exists that the 90nm Power5 derivatives still could hit major snags. We'll know by the end of May - and I'm sure every Mac rumor and news site will have updates on that.
  • Reply 44 of 148
    Quote:

    Originally posted by Mr. MacPhisto

    So, I know nothing about PCIe implementation, although I hope it will come with the next revision. It is my guess that revision will come at WWDC - just because of the one year deadline Jobs placed on Apple.



    From what I've pieced together, however, it is my belief that Jobs gave that date as a conservative estimate, hoping that IBM would deliver 90nm 970FXs at 3GHZ for introduction in January and shipment sometime in February.




    Thanks for the info. I'm glad someone else feels Steve's original 3GHz estimate was conservative. At the time I was absolutely floored that he would discuss the roadmap at all, let alone put up a giant slide with a specific clockspeed and date. I remember thinking "wow, even his Plan B must be for 3GHz or else he never would open his mouth like this."



    Well, we're definitely on Plan B now so let's hope it always allowed for 3GHz too.
  • Reply 45 of 148
    pbpb Posts: 4,255member
    Quote:

    Originally posted by Mr. MacPhisto

    Apparently, Apple is reluctant to cut ties with Moto. They like some of the chips they are presently working on and have regained confidence in them since the semiconductor division split was announced - coupled with the opening of the Crolles fab. IBMers in Fishkill have heard that Apple expects Moto to hit 90nm this summer in quantity - coupled with some other processor improvements. As to what they are, I don't know. I would guess we'd see a L2 cache increase.



    Are you talking about these improvements on the G4 (look at the end of the pdf)? I only hope they manage to deliver in time; it would be a very decent mobile chip (single or dual core) for the Powerbooks.
  • Reply 46 of 148
    snoopysnoopy Posts: 1,901member
    Quote:

    Originally posted by Mr. MacPhisto





    . . . I've been told there will be chips capable of being placed in notebooks - but don't know if Apple will use them or not.






    Assuming Power 5 derivative and 970FX are both built with essentially the same 90 nm process, the 970FX should run cooler, no? With smaller L2 cache and no SMT the device count is less and power dissipation should be lower, at the same clock rate. Possibly IBM made a design breakthrough and runs the Power 5 derivative at lower voltage, but likely not. My guess is that the 970FX will be the lower power chip.



    Since IBM appears to have no worries about the Power 5 derivative, it lends support for the rumor theory that the 970FX has a packaging problem. With a much smaller chip size, IBM and Apple may have decided to shrink the package, to save space and/or cut costs. If the package fails at higher power, it would restrict the upper clock rate. In the same package as the 970, the 970FX likely does 3 GHz. IBM may be doing a long burn in on these parts, to be sure the package stays intact at 2 GHz. This would really slow things down. It seems to fit what we see happening. The package can be fixed, but it takes a little time.



    Regarding PowerBooks and the Power 5 derivative, IBM quoted low power for the 970 too, at sufficiently reduced clock rate and supply voltage. It is a characteristic of the whole family of chips it seems. I believe the 970FX will become the low power champ, however.
  • Reply 47 of 148
    kurtkurt Posts: 225member
    Quote:

    Originally posted by Ensign Pulver

    Thanks for the info. I'm glad someone else feels Steve's original 3GHz estimate was conservative. At the time I was absolutely floored that he would discuss the roadmap at all, let alone put up a giant slide with a specific clockspeed and date. I remember thinking "wow, even his Plan B must be for 3GHz or else he never would open his mouth like this."



    Well, we're definitely on Plan B now so let's hope it always allowed for 3GHz too.




    With the recent problems IBM has had with the 90 nanometer process, it is easy to question whether Apple would be able to deliver 3GHz in the original time frame. (June or September depending on whether he meant one year from announcement or one year from shipment.) But I would have thought that Apple would have made some announcement in their recent analyst call or just a caution to analysts in general if they do not expect to make it. That would affect the stock price since it is expected to get to the faster machines and people would question why it was not happening.



    Confirmation but lack of evidence to the contrary? What are the exact rules concerning what they need to divulge about items that could affect the stock price?
  • Reply 48 of 148
    programmerprogrammer Posts: 3,467member
    Quote:

    Originally posted by Mr. MacPhisto

    As for the 65nm process at IBM. They seem to feel that the transition will be easier than the 90nm transition has been. The hope is that 65nm chips could be fabbed by the end of the year. Fishkill is ready to go @ 65nm and much of the technology has been implemented at 90nm.



    They think "90 -> 65" will be easier than "130 -> 90", not that "130 -> 65" will be easier than "130 -> 90". The still need to prove out a production run or two on 90 nm. This summer seems a reasonable timeframe for that to hit the market though.



    There are a few companies with big plans riding on IBM's achieving the 65 nm node next year. The evidence that they can achieve this must be pretty compelling. My guess is that Apple's recent comments about a PBG5 not until next year hinge on a 65nm 970 variant.



    With the fairly rapid timeline of 90 -> 65, we should see a chip (typically called the 975) show up between now in then in an Apple product cycle or two. If they're aiming at two cycles per chip (intro & then bump) then we saw the 970 intro last WWDC, they missed the bump early this year due to problems, and they ought to be due for a new chip this summer or fall in order to fullfill Steve's promise and then a speed bump of that one in the spring, with the new 65 nm chip next summer or fall.
  • Reply 49 of 148
    programmerprogrammer Posts: 3,467member
    Quote:

    Originally posted by Kurt

    With the recent problems IBM has had with the 90 nanometer process, it is easy to question whether Apple would be able to deliver 3GHz in the original time frame. (June or September depending on whether he meant one year from announcement or one year from shipment.)



    Keep in mind that Mr. Jobs went through the 500 -> 450 debacle. It was he who had egg on his face over that one, and he was extremely unhappy about it. For him to come out and say publicly that Apple would hit 3 GHz by end-of-summer 2004, he would have had to have a very high degree of confidence in achieving that. Not only would they have needed to have 3 GHz running in their labs already, but they would need to have PlanA, PlanB, and maybe even a PlanC. The engineers implementing those plans would be very aware of the noose around their collective necks if they promised it but failed to deliver. All other plans would be secondary -- including the spring PowerMac revision.
  • Reply 50 of 148
    Quote:

    [Fixed the title -Amorph]



    Ok, I just noticed you fixed this for me Amorph.



    Thank You Thank You thank You.



    OMF
  • Reply 51 of 148
    kurtkurt Posts: 225member
    Quote:

    Originally posted by Programmer

    Keep in mind that Mr. Jobs went through the 500 -> 450 debacle. It was he who had egg on his face over that one, and he was extremely unhappy about it. For him to come out and say publicly that Apple would hit 3 GHz by end-of-summer 2004, he would have had to have a very high degree of confidence in achieving that. Not only would they have needed to have 3 GHz running in their labs already, but they would need to have PlanA, PlanB, and maybe even a PlanC. The engineers implementing those plans would be very aware of the noose around their collective necks if they promised it but failed to deliver. All other plans would be secondary -- including the spring PowerMac revision.



    I agree but my point was isn't there some FTC rules that state Apple must provide news that may affect their earnings? Things like a major increase in costs or a loss of a critical supplier. I would think that the inability to get to 3 GHz by summer 2004 would fall into that category. Also, they are downplaying the G5 Powerbooks pretty heavily. Therefore, does the lack of announcements that we will not be getting to 3GHz by end of summer indicate we will and things are unchanged?
  • Reply 52 of 148
    geekmeetgeekmeet Posts: 107member
    Quote:

    Originally posted by Mr. MacPhisto

    The Power5 derivative should be able to run in the PowerBooks and the PowerMacs. After it moves over to 65nm a new dual-core chip should become available, coupled with multi-threading.



    I have a feeling that FreeScale (the spunoff Moto semiconductors unit) may be able to provide some impressive chips for the consumer market in the near future. The people in IBM are well aware of them and consider them to be very serious competition in the embedded market - but are aware that Apple also is impressed with proposed future offerings. I can't provide anything firm because I don't know anyone over there - but persistant rumors indicate they will move to 90nm this summer and there's a 7400 series dual core chip in the works. I hope they've got something good in the near future, but I won't hold my breath. The cancellation of the VX project (which could have been fabbed in volume in March - though the 90nm problems would have delayed that as well) indicates to me that Apple has some kind of good reason to continue relations with Moto/FreeScale. Apple spent millions developing that chip with IBM. Some of that development will go into the system-on-a-chip project that IBM is working on for Apple. Even so, that's a lot of time and money for something that won't see regular production.




    well well well.....................

    MOTOROLA BACK?

    why would apple not go through with deploying a processor that they in large part payed for unless motorola had something pretty COMPELLING coming.



    very interesting.

    all you motorola haters completely wrote moto off and look who has egg all over their face.



    apple IS going back to motorola beacause of one thing:the industry is trending towards smaller,faster and less power and ONLY motorola can deliver that.



    ITS ALL ABOUT LAPTOPS SILLY!

    and portables!



    the industry is trending that way and apple HAS to trend that way with its processors.



    NOW



    mr.macphisto



    i have a few questions for you.

    is it your gut feeling that apple will announce the 975 chip at wwdc?

    also.........since the "vx" chips have been cancelled when will motorola release more compelling chips?

    im assuming very soon since the purported "vx" chip could have been here already.



    and whats this "system on a chip" thing you spoke about?

    is that apples sytem controller chip?



    long live motorola!!

    many of you think you know.....but you dont know..........and you never will either!



  • Reply 53 of 148
    amorphamorph Posts: 7,112member
    Quote:

    Originally posted by geekmeet

    also.........since the "vx" chips have been cancelled when will motorola release more compelling chips?



    The stuff about "90nm this summer" is actually from Moto/FreeScale employees and management. They've also been hinting at big things this year, starting this summer.



    Quote:

    and whats this "system on a chip" thing you spoke about?

    is that apples sytem controller chip?




    "System on chip" means a lot of things, but in this context it's taken to mean a chip that contains a CPU along with some other functionality. A 7457 variant with an on-die memory controller would be a SoC, technically.



    SoC designs are mostly used to cut costs - the more functionality you can stuff onto the fewest chips, the cheaper your computer is. The Holy Grail is a single chip with a CPU, GPU, and all controller functionality, so that the motherboard becomes a mere vestige. In the case of moving the memory controller on board, there are also possible performance advantages.



    Quote:

    many of you think you know.....but you dont know..........and you never will either!



    So how long have you been working for Mot?
  • Reply 54 of 148
    matsumatsu Posts: 6,558member
    I wonder if moto/freescale will have a 64 bit variant suitable for laptop use. It isn't critical at this time. There ain't no 64 bit "mobile" chips from intel or AMD. There are 64 bit "luggables" based on AMD chips but nothing we'd call a true "laptop".



    I ask because if there is something with that huge address space, we might never see an IBM based Powerbook at all, what would be the point, so long as the moto offering is small/cool/frugal, and offers industry standard level performance. If there's no need for 64 bits, or there's a way to cover that need with a more mobile friendly chip, then there's no need for IBM in a laptop.



    If, however, there is a pressing need for a mobile professional 64 bit platform, and no small/cool solution, I do see that we may have a third tire of Apple laptops added.



    iBook G4

    Powerbook G4+ ?

    Powerbook G5



    All to exist simultaneously, 32 bit machines with an emphasis on battery life, 64 bit machines with an emphasis on number crunching.
  • Reply 55 of 148
    kurtkurt Posts: 225member
    Just curious. Has there been any official announcements or even hints that there is any PowerPCs beyond the 970fx. I know IBM does somewhere mention it as the '970 family' or something like that. I don't remember seeing anything more concrete than that. Is this just speculation (albeit probably true) that IBM will improve the chip to stay competitive?
  • Reply 56 of 148
    programmerprogrammer Posts: 3,467member
    Quote:

    Originally posted by Kurt

    Just curious. Has there been any official announcements or even hints that there is any PowerPCs beyond the 970fx. I know IBM does somewhere mention it as the '970 family' or something like that. I don't remember seeing anything more concrete than that. Is this just speculation (albeit probably true) that IBM will improve the chip to stay competitive?



    They've changed it now, but their website used to have a pulled-down menu labelled "9xx Processor Family". There have been various other hints.







    System-on-chip (SoC) is very important in embedded and portable applications. Fewer chips means smaller boards, lower power/heat, lower cost, and typically higher performance per unit power/heat. If Freescale can do one thing right it would be to build a G4 7447A for Apple with (at least) a memory controller built-in. Doing it on Crolles 90 nm process would be even better. Adding other devices into the SoC would be icing on the cake (e.g. ethernet, wireless, AGP, PCI, etc). They were working on these technologies for their 85xx chips, but those appeared to die or stall due to their fab issues. Bumping the clock rate at that point would be less important than the drop in power/heat... a 1.5 GHz G4++ like this would perform substantially better than the current 1.5 GHz G4 and would exhibit much longer battery life and run cooler in the case.
  • Reply 57 of 148
    henriokhenriok Posts: 537member
    Quote:

    Originally posted by Programmer

    Not only would they have needed to have 3 GHz running in their labs already, but they would need to have PlanA, PlanB, and maybe even a PlanC.



    I hope PlanC doesn't involve names like Pentium, but that would get us beyond 3 GHz... \
  • Reply 58 of 148
    tfworldtfworld Posts: 181member
    To what? 3.4Ghz?
  • Reply 59 of 148
    Quote:

    Originally posted by geekmeet

    mr.macphisto



    i have a few questions for you.

    is it your gut feeling that apple will announce the 975 chip at wwdc?

    also.........since the "vx" chips have been cancelled when will motorola release more compelling chips?

    im assuming very soon since the purported "vx" chip could have been here already.



    and whats this "system on a chip" thing you spoke about?

    is that apples sytem controller chip?



    long live motorola!!

    many of you think you know.....but you dont know..........and you never will either!







    My gut says that Apple will announce a PowerMac based on the Power5 derivative (commonly called the 975). If IBM's production schedule holds, these chips should be available in quantity during the summer.



    The VX chips could not have been here already because of the snags @90nm. They were designed to be fabbed at 90nm, so their timetable was pushed back to late summer when they were canned in January. I would assume that FreeScale has got something big coming this summer. Amorph has tackled all this stuff better than I could - but Moto's own documents indicate a possible 2GHZ G4 with faster bus, increased cache, and better memory support. I'm not betting the farm, but if this happens then the G4 becomes a very viable chip once more. The Moto stuff I've read inicates that FreeScale thinks they can take the 74xx up to 3GHZ. Don't know anything about what's happening at FreeScale, but I do think Apple expects some chips from them that are superior to what they worked on with IBM (the VX supported up to a quad-pumped bus (4x100MHZ). However, my guess would be that we won't see these things in 'Books until September - though that's a hunch based on length of revisions. The could surprise me and revise them in the middle of the summer, right before the back-to-school season. This would be a welcome surprise - though I'm sure it'd tick off those who are buying the new laptops.



    SOC was tackled by Amorph. In theory, SOC should cut cost and often power consumption. This would also make the logic board easier and cheaper to design. I'm not sure if FreeScale has got any SOC plans of their own that Apple could use.
  • Reply 60 of 148
    Is not SOC a specialty of MOTO-FreeScale?
Sign In or Register to comment.