New G5 chips? 4 flavors: 2GHz, 2.2GHz, 2.4GHz, and 2.6GHz

Posted:
in Future Apple Hardware edited January 2014
IBM, as recently as this week, has begun the process of fabricating 90nm G5s in volume. According to sources, test yields have shown the revision to clock up to 2.6GHz and come in a variety-pack of 4 flavors: 2GHz, 2.2GHz, 2.4GHz, and 2.6GHz. Additionally, yields of the new 90nm microprocessor are going well.



What do you think the G5 line will be with this information? Will there be 4 G5 in the line or 3? Does this mean a possible new 90nm 2ghz could go into a imac or powerbook?



Will the G5 line be...



Single 2.2, Dual 2.4, Dual 2.6?
«13456

Comments

  • Reply 1 of 117
    Yeah, I was also wondering about that. I would bet that Apple sticks with three PowerMacs. That means either that those 2 GHZ models are going into the iMacs or IBM servers. I think it's still too soon for G5 PowerBooks.
  • Reply 2 of 117
    dhagan4755dhagan4755 Posts: 2,150member
    Any word on the power consumption of these new G5 chips?
  • Reply 3 of 117
    dmband0026dmband0026 Posts: 2,345member
    Quote:

    Originally posted by DHagan4755

    Any word on the power consumption of these new G5 chips?



    Less than the 130nm process they are built on now. The 90nm is supposed to consume considerably less power than the 130 that we have now.
  • Reply 4 of 117
    placeboplacebo Posts: 5,767member
    Go Apple!?
  • Reply 5 of 117
    nr9nr9 Posts: 182member
    In other news, Intel has increased their Pentium 4 speed to 3.4GHz. They are expected to reach 4.0GHz next year.
  • Reply 6 of 117
    Quote:

    Originally posted by Placenta

    Go Apple!?









    ?
  • Reply 7 of 117
    jubelumjubelum Posts: 4,490member
    Nr9 has a point... Intel is still winning the Mhz war... in raw numbers, but let's look at the roadmap for the 970. We might not be there yet in simple Mhz, but the G5 architecture is brilliant. Apple's 18 month outlook is better than it has ever been.



    IMHO, a 2.6 dual would take care of me for at least 2-3 years. Smart consumers know that you should never buy the first release of anything that is redesigned (especially cars, etc) because the real-world kinks have not emerged. Rev B is the one for me.



    Just wondering what people have been hearing in terms of Mhz from possible switchers. I could not get any of my PC friends to even LOOK at a G4 based simply on the Mhz myth. Now that gap is narrowing faster than ever. Anyone have an opinion on how the new G5's are playing with switchers, now that Ghz are getting competitive in numbers?



    Also, as for the heat/power issue... are we now permanently consigned to have monster heat sinks on everything? How much will the 90nm help with the heat issue for smaller enclosures?
  • Reply 8 of 117
    nr9nr9 Posts: 182member
    Quote:

    Originally posted by Jubelum

    Nr9 has a point... Intel is still winning the Mhz war... in raw numbers, but let's look at the roadmap for the 970. We might not be there yet in simple Mhz, but the G5 architecture is brilliant. Apple's 18 month outlook is better than it has ever been.



    everytime i see "Nr9 has a point" i laugh.
  • Reply 9 of 117
    So if Apple reaches 3GHz next year, and Intel 4GHz, the gap in absolute clock speed will remain the same, and the percentage gap will decrease.



    Sounds like good news for Apple to me
  • Reply 10 of 117
    Where does this information come from? There's no attribution whatsoever. EDIT: Answered my own question, NM
  • Reply 11 of 117
    kid kkid k Posts: 15member
    does anyone know how sheer transistor numbers aboard the chip effect speed? heres is where i truly give credit to IBM.. why? i read that they produced nearly the same results with a little more than 1/3 (as i recall: it may be closer to 1/2.. not sure) as many transistors as the opterons/ fx/amd64.. The opterons did however.. win. BUT, something has to be said for that. its one of the reason i switched.. What if they were equal....???



    In my mind, for the future, IBM has a clear edge here.. eventually space will run out as materials reach their limits..at that point it will be who utilizes their transistors the best.. along with architecture outside that of course... but, lets face it, unless you're rendering like pixar (workstation vs. home use), current cpu speeds are even beyond the human ability to notice major differences under normal use that is (i.e. internet.. music, etc.).. we'll see what happens.
  • Reply 12 of 117
    mmmpiemmmpie Posts: 628member
    I think you'll find that those x86 with lots of transistors have much bigger caches than the 970. The bit about x86 that used to be tuff ( instruction decoding ) is now a well solved problem, and no longer uses a large percentage of available chip space. Hence it is pretty much irrelevant. You can assume that for a well designed chip, more transistors are being used for useful stuff. Cache. more functional units, smt, dual core, integrated memory controllers etc etc.



    The reality is, Intel are good at building chips, AMD are pretty good to. IBM are also doing well. Nobody is taking huge strides out into the front. The conceptual differences between cpu's dont really add up to much, except perhaps design difficulty ( see Itanic ).



    As long as IBM remain commited Apple will have a competitive cpu to use. IBM have good tools, so that might be enough to overcome the small number of systems that Apple is shipping.
  • Reply 13 of 117
    jlljll Posts: 2,709member
    Quote:

    Originally posted by Nr9

    In other news, Intel has increased their Pentium 4 speed to 3.4GHz. They are expected to reach 4.0GHz next year.



    Yeah, Intel added 6.25% clock speed in a year while IBM added 30% in three months





    Quote:

    Originally posted by crayz

    So if Apple reaches 3GHz next year, and Intel 4GHz, the gap in absolute clock speed will remain the same, and the percentage gap will decrease.



    Apple (IBM) is expected to reach 3GHz next summer and it seems they are ahead of schedule.



    Perhaps Intel's 4GHz Pentium will be here this time next year?
  • Reply 14 of 117
    kroehlkroehl Posts: 164member
    Quote:

    Originally posted by JLL

    Yeah, Intel added 6.25% clock speed in a year while IBM added 30% in three months



    Apple (IBM) is expected to reach 3GHz next summer and it seems they are ahead of schedule.



    Perhaps Intel's 4GHz Pentium will be here this time next year?




    Apple is already shipping a 4GHz system (2x2). If, by summer, they'll have a 2x3GHz system on the shelves, I think it'll be time to put the stupid megahurtz d!ckwaving competition to bed (no pun intended).



    Noone should count Intel out of the game though. What we're seeing is good competition and a huge catch-up leap by IBM, but it's not like someone just brought an M1 tank to the battle of Waterloo on their time machine.
  • Reply 15 of 117
    First let me state, I no nothign more than what I think I read.



    Quote:

    Originally posted by mmmpie

    [B]I think you'll find that those x86 with lots of transistors have much bigger caches than the 970./B]



    The L3 cache does make a huge difference:



    Quote:

    The original Northwood core is made up of 55 million transistors. Adding 2 MegaBytes of cache, where each bit requires six transistors, to the die adds 100.7 million transistors for the SRAM cells alone. This does not include the interface logic that will take up as much as another 25 % transistor count overhead, and by the end of the equation, we are looking at the proud number of approximately 125 million transistors, only for the L3 cache - more than twice the number of transistors in the Northwood core. The total transistor count for the new processor would be ~ 180 million transistors, the number confirmed by Intel is 178 million.



    A 970 has 52 million transistors. The old P4 (non EE) had only 55 million. So as mmmpie was getting at, the difference in transistor count isn't really that great. The real differnce is in heat (on the EE P4 or the older one).
  • Reply 16 of 117
    Quote:

    Originally posted by kupan787

    First let me state, I no nothign more than what I think I read.



    The L3 cache does make a huge difference:



    A 970 has 52 million transistors. The old P4 (non EE) had only 55 million. So as mmmpie was getting at, the difference in transistor count isn't really that great. The real differnce is in heat (on the EE P4 or the older one).




    Well the L3 makes a big difference when your FSB+Memory is slower, but will get in the way if the FSB+Memory is faster than a L3 as would be the case with the 970. So the L3 is simply not needed and would only slow things down on 970 with its current FSB. The next step for IBM/Apple would be to double the L2 which runs at core speed.



    Transistor count is very important as the more transistors the greater the resistance which means that you have to run the chip at a higher voltage. The relationship between heat and voltage is quadratic and thus serves the highest heat penalty. and upping the frequency at the same voltage results in a linear heat increase. This is one of the areas that Intels design is really hindered, Prescot even at 90NM is reported to need 140W, so you see Intels problem even taking Prescot from 3.4 to 4GHz.



    IBM has solved the problem though with their PowerTune technology which allows them to seperate different parts of the processor into Voltage Islands, allowing the core processor to run at a higher voltage while other parts are running at a lower voltage. Too, their dynamic power management not only adjusts the frequency but also the voltage.
  • Reply 17 of 117
    Hummm.... power increase... (Homer Simson Style)
  • Reply 18 of 117
    henriokhenriok Posts: 537member
    Quote:

    Originally posted by Nr9

    In other news, Intel has increased their Pentium 4 speed to 3.4GHz. They are expected to reach 4.0GHz next year.



    Intel was expeceted to release Prescott this fall, but they are delayed until february, at the earliest. Their roadmaps show 4 GHz Q4'04, and if that date is delayed as much as Prescott's date is then they won't reach 4 GHz in 2004 but 2005. At the pace IBM is accelerating their chips, they too will reach 4 GHz in 2005. If we're a bit lucky and Intel is unlucky.. we'll both reach 4 GHz in 2005. Pretty neat.



    And.. A 3.4 GHz Pentium 4 EE costs 999 dollars. That's quite pricy. And.. I assume that it drwas quite a lot of power to. Prescott doesn't seem to change that.
  • Reply 19 of 117
    tak1108tak1108 Posts: 222member
    Single 2Ghz

    Dual 2.4 Ghz

    Dual 2.6Ghz



    Single 2.2Ghz Xserve

    Dual 2.2Ghz Xserve







    Please, Please, Please,
  • Reply 20 of 117
    An interesting (and nice to have) problem is starting to arise for Apple with the rapid scaling of the G5. Right now the difference in processor speed between the mid-level and top machine is 11%. $500 gives you that gain plus an 11% faster front side bus and a better graphics card.



    If Apple goes to a 2.2 SP, 2.4 DP, 2.6 DP line-up, they?re now asking for the big bucks for a speed increase of only 8%. A lot of folks might say, ?Why pay, is it really worth 20% more for an 8% speed increase??



    Therefore, I?m hoping that the line-up might look a little more like this:



    PowerMac:

    Single 2.0 GHz $1,699

    Dual 2.2 GHz $2,399

    Dual 2.6 GHz $2,999



    Xserve

    Single 2.2 GHz $2,499

    Dual 2.2 GHz $3,499



    New iMac

    Single 2.0 GHz $1,099, $1,599, $1,999 for 15?, 17? and 20? (Okay, they?ll probably be higher but ya gotta have dreams))
Sign In or Register to comment.