Developer Notes points to possible reason for delay

kupan787kupan787
Posted:
in Future Apple Hardware edited January 2014
Apple has posted the G5 Xserve Developer Notes . One interesting tidbit is that the system controller is 130 nm, not 90 nm as widely rumoured.



Click on "Architecture", and then on "U3H Bridge and Memory Controller". It is near the bottom in the "Processor Bus" section.



Quote:

The Xserve G5 system controller is built with state-of-the-art 130-nanometer, SOI technology.



Could this be the reason for the delay of PowerMacs? Since the Xserves are only using 2GHz parts, a new SC isn't really needed. But for faster chips, if they are having problems getting the system control on a 90nm process, it would make sense.

Comments

  • Reply 1 of 11
    @homenow@homenow Posts: 998member
    Why would the system controller need to be on 9 nm? It just needs to be able to communicate withe the 970 and the system at a given speed, currently 1/2 the clock speed of the processor.
  • Reply 2 of 11
    crusadercrusader Posts: 1,129member
    Yea, I don't think that the system controller has to be built using the same process as the chip it's communicating with. For a 3 Ghz G5 the system controller would only have to run at 1.5 Ghz. Holy crap the system controller will run at a higher speed than the G4 chips. That will be an excellent day indeed.



    BTW: Thanks for the linkage to the Developer Notes.
  • Reply 3 of 11
    kupan787kupan787 Posts: 586member
    Quote:

    Originally posted by Crusader

    Yea, I don't think that the system controller has to be built using the same process as the chip it's communicating with.



    That is a given since the Xserve are 90nm 970s but the system control is 130nm.



    Quote:

    For a 3 Ghz G5 the system controller would only have to run at 1.5 Ghz. Holy crap the system controller will run at a higher speed than the G4 chips. That will be an excellent day indeed.



    BTW: Thanks for the linkage to the Developer Notes.



    There was a rumor going round that the current system control could only clock up to 1.1GHz. Do we know with certantity that the current system control can clock up to 1.5Ghz, and still run at a reasonable heat level?
  • Reply 4 of 11
    oldmacfanoldmacfan Posts: 501member
    Hmm,

    Where we not debating and pondering an article about IBM producing a 90nm soc two months ago?



    Could this be the same chip?



    Just pondering, please correct me if I am wrong.
  • Reply 5 of 11
    kupan787kupan787 Posts: 586member
    Quote:

    Originally posted by oldmacfan

    Hmm,

    Where we not debating and pondering an article about IBM producing a 90nm soc two months ago?



    Could this be the same chip?



    Just pondering, please correct me if I am wrong.



    If I am not mistaken a SOC (system on a chip) and an SC (system controler) are two different beasts. So the fact that IBM is making 90nm SOC and 90nm 970s doesn't necessarily mean that they aren't having problems making 90nm system controlers for the 970.
  • Reply 6 of 11
    programmerprogrammer Posts: 3,458member
    The fact that the system controller is a new chip is enough reason to explain the delays. If Apple is having troubles with a new design or version of their chip (perhaps one that is supposed to run cooler, support PowerTune, or have any fancy new features) then this would slip their schedule just as surely as the 970FX having yield problems. In fact, based on the info IBM revealed about success rates at Fishkill, it would seem much more likely that the problem lies with the Apple chip and not the 970FX (which is virtually unchanged from the 970).
  • Reply 7 of 11
    hasapihasapi Posts: 290member
    Quote:

    Originally posted by kupan787

    Apple has posted the G5 Xserve Developer Notes . One interesting tidbit is that the system controller is 130 nm, not 90 nm as widely rumoured.



    Click on "Architecture", and then on "U3H Bridge and Memory Controller". It is near the bottom in the "Processor Bus" section.







    Could this be the reason for the delay of PowerMacs? Since the Xserves are only using 2GHz parts, a new SC isn't really needed. But for faster chips, if they are having problems getting the system control on a 90nm process, it would make sense.



    Well picked up kupan! it was ASSUMED that the SC chip would be fabbed with the CPU, of course the PMac nor the xServe require the new SC, as for a G5 Powerbook?, definetly.



    Programmers explanation is more plausible than the spurious notions being thrown around?.
  • Reply 8 of 11
    rickagrickag Posts: 1,626member
    Quote:

    Originally posted by oldmacfan

    Hmm,

    Where we not debating and pondering an article about IBM producing a 90nm soc two months ago?



    Could this be the same chip?



    Just pondering, please correct me if I am wrong.



    Yes we were. The IBM rep was quoted as saying they had just begun shipping 0.09µm SOC chips to Apple. So either



    the rep was misquoted and meant SC or

    the rep made a mistake during the interview and meant SC or

    the rep was using industry accepted terminology(??) for SOC = SC or

    Apple did order SOC processors for an undisclosed product or

    the rep made a mistake and meant 0.13µm SOC or

    the rep was misquoted and said 0.13µm SC or

    or or or



    pick your poison
  • Reply 9 of 11
    bigcbigc Posts: 1,224member
    Keyword seems to be misquoted...
  • Reply 10 of 11
    rickagrickag Posts: 1,626member
    Quote:

    Originally posted by Bigc

    Keyword seems to be misquoted...



    Which one? 0.13µm or SOC?



    If SOC, then why the 0.13µm SC in the xServe?

    If 0.13µm then where's the SOC chip going?

    If both SOC and 0.13µm then the journalist should be required to go back to school.

    this dissecting of the IBM quote could go on until WWDC, couldn't it? After all we need something to discuss with so many delays.
  • Reply 11 of 11
    bigcbigc Posts: 1,224member
    Well I don't know, most of the reporters I see on TV look like their mommy dressed them...as to the ones that report for Web distribution, my guess is that there looking fo their mommy (i.e., somewhat lost).
Sign In or Register to comment.