Cell details

12346

Comments

  • Reply 101 of 134
    amorphamorph Posts: 7,112member
    Quote:

    Originally posted by Brendon

    Are you only referring to Cell? Because I thought that the 970 was SOI.



    Strained Silicon Directly On Insulator is an enhancement to SOI.
  • Reply 102 of 134
    hirohiro Posts: 2,663member
    Quote:

    Originally posted by Brendon

    I believe that we were talking about a daughter card, maybe a big daughter card, but still. What would the size be? Oh 8.7inches squared @90nm, or 4.5inches squared @65nm.



    At 65nm daughter card is no problem, but 90nm chip is huge and I would think that supporting it and cooling it would be difficult unless you have this water cooling thing down




    There is a difference between 221mm squared and 221 sq.mm. See the pic? Not an 8"x8" square (~41,300 sq.mm). ~10mm x ~20mm. Just barely over 3/4 of an inch on the long side. That does not require any kind of huge daughtercard to support it.



  • Reply 103 of 134
    Quote:

    Originally posted by Hiro

    There is a difference between 221mm squared and 221 sq.mm. See the pic? Not an 8"x8" square (~41,300 sq.mm). ~10mm x ~20mm. Just barely over 3/4 of an inch on the long side. That does not require any kind of huge daughtercard to support it.







    Thanks for the closeup of that image, yea I think that would fit in nicely. My bad Also they would had to have figured out thow to manufacture this with the hole in the middle because I think that the size I quoted is bigger than the wafer.
  • Reply 104 of 134
    amorphamorph Posts: 7,112member
    Actually, I heard that it is 221mm squared, and IBM found someone 50 feet tall to hold it for the camera.
  • Reply 105 of 134
    powerdocpowerdoc Posts: 8,123member
    Quote:

    Originally posted by Programmer

    Sure, it might cost you half your clock rate. Oh, right, that's about where we are now anyhow...



    Right, but when the cell will be released, where we will be ?



    Will you take a dual core Power PC over this Cell chip for an Apple desktop computer ?
  • Reply 106 of 134
    thttht Posts: 5,681member
    Quote:

    Originally posted by Amorph

    I never doubted that. The G4 has full double-precision FP support, too. But the 970 has two full FP units. The question was not, can Cell do 64-bit FP, but can it muster the performance of a 970 at 64-bit FP. (See below before answering.)



    RealWorldTech said the SPE performs DP FPU ops at 10% of the performance of single precision ops. Looks like there is or the equivalent of a 10 cycle latency for double precision. If all 8 SPEs are going at the same time, it may be good, but that's a big if.
  • Reply 107 of 134
    thttht Posts: 5,681member
    Quote:

    Originally posted by rickag

    Speaking of 100+ watts, I wonder how far along IBM is with their Strained Silicon Directly on Insulator technology? Heck, for that matter, I wonder how far along they are in implementing SOI, the last mention of it was Chipwork's statement that they were expecting samples "soon", whatever that means?



    No word on SSDOI, but there were a few mentions that Cell will be produced at 90 nm with SOI and low-k. Maybe we'll see a Powerbook G5 in the Summer afterall.
  • Reply 108 of 134
    rickagrickag Posts: 1,626member
    Quote:

    Originally posted by Brendon

    Are you only referring to Cell? Because I thought that the 970 was SOI.



    You're right. I meant to say low-k dialectric, but I never know what I'm talking about and on top of that I think I'm getting senile in my old age.
  • Reply 109 of 134
    Quote:

    Originally posted by rickag

    You're right. I meant to say low-k dialectric, but I never know what I'm talking about and on top of that I think I'm getting senile in my old age.



    Really, I just found out that my house is allot smaller than I thought it was It was 2300 feet squared now I learn that it is only 2300 square feet.



    Thanks Hiro
  • Reply 110 of 134
    onlookeronlooker Posts: 5,252member
    Quote:

    Originally posted by Bigc

    hmm, which would be better; A dual processor dual core set-up or a single processor dual-core with cell attributes attached...DP dual core sounds a little redundant...



    Redundant. maybe to some, but I think it sounds like a machine that could turn out excellent rendering times. Like those high $$$ Quad Opteron motherboards, but cheaper. (hopefully cheaper)



    I figure Apple has to have DP dual core motherboards when the dual core chips come. Nvidia's nForce Pro for Opteron ( in which DP version is the popular version) is supposed to be capable of holding current Opteron processors, and Dual core Opteron Processors according to nvidia. Once the chips are released all an nForce owner has to do is buy the processors. If your on the x86 side, that's it. Pretty cool huh. I wish we could do that. Save on some serious upgrade costs.

    Nevertheless the rest of the motherboard manufacturers are still going to be making 4 way versions for the Opteron, and when they get dual core that will make those current 4 way mobo's into theoretical 8 way's.

    So, excluding those 8 way versions. If Apple reduces to one processor slot for the PowerMac it will be like the current Opteron 2 way motherboards are getting a lift to where those high $$$ 4 way motherboards are, and our PowerMac will remain at the same level it currently sits at. That is not building my confidence.
  • Reply 111 of 134
    kurtkurt Posts: 225member
    Quote:

    Originally posted by Powerdoc

    Right, but when the cell will be released, where we will be ?



    Will you take a dual core Power PC over this Cell chip for an Apple desktop computer ?




    Maybe I missed the discussion on this already but, is it possible/advantageous to use this chip in concert with a 970? Somewhat similar to the old QuadraAVs that had a separate DSP chip. Programmer mentioned how this is similar to a GPU and Apple is doing alot with its GPUs lately in Quartz extreme and its professional video software. That might be where this chip shines the most.



    Of course there is al that speculation that Apple will provide software to Sony for upcoming machines. Could this be what Steve is talking about in that recent Fortune article?



    It is fun to surmise.
  • Reply 112 of 134
    powerdocpowerdoc Posts: 8,123member
    Quote:

    Originally posted by Kurt

    Maybe I missed the discussion on this already but, is it possible/advantageous to use this chip in concert with a 970? Somewhat similar to the old QuadraAVs that had a separate DSP chip. Programmer mentioned how this is similar to a GPU and Apple is doing alot with its GPUs lately in Quartz extreme and its professional video software. That might be where this chip shines the most.



    Of course there is al that speculation that Apple will provide software to Sony for upcoming machines. Could this be what Steve is talking about in that recent Fortune article?



    It is fun to surmise.




    Well Programmer said that Cell is not simply the ship presented here, but a whole new architecture concept. It's possible to see many more variants, and why not a custom variant for an Apple desktop.



    I just say that I don't think that the chip presented here, will be the best choice for an Apple desktop. I think that for a desktop computer, you can use in a more efficient way 220 millions of transistors (the current G5 is under 60 millions transistors).
  • Reply 113 of 134
    mattyjmattyj Posts: 898member
    I feel as more emphasis should be put not on the cell chip itself, but rather the problem of producing the right code for it.



    I have been told that as far as programming goes now, it is chaotic and not very organised, or could at least be a lot 'neater'. The difficulty with a new architecture is getting the programmers to code properly for it. The PS3 is going to use this chip, so ports of PS3 games to the Xbox 2 may become far more difficult than it is from the current PS2 to Xbox (although we have middleware such as renderware).



    Programmer perhaps you could enlighten me upon the subject.
  • Reply 114 of 134
    ompusompus Posts: 163member
    Virtually no-one has actually programmed for this chip. What tricks IBM and team have figured out, and precisely how difficult it will be to use the cell, are intriguing but open questions.
  • Reply 115 of 134
    powerdocpowerdoc Posts: 8,123member
    Quote:

    Originally posted by mattyj

    I feel as more emphasis should be put not on the cell chip itself, but rather the problem of producing the right code for it.



    I have been told that as far as programming goes now, it is chaotic and not very organised, or could at least be a lot 'neater'. The difficulty with a new architecture is getting the programmers to code properly for it. The PS3 is going to use this chip, so ports of PS3 games to the Xbox 2 may become far more difficult than it is from the current PS2 to Xbox (although we have middleware such as renderware).



    Programmer perhaps you could enlighten me upon the subject.




    In the case of the PS 3, I have no doubt that the programmers will do the right job to take advantage of this new beast.

    This is an another story to ask to companies like Adobe, to invest time and money for optmising code. Adobe like update, when it bring them back money. The same is true for all programmers.
  • Reply 116 of 134
    mattyjmattyj Posts: 898member
    Well the problem is, with the exception of the companies that make the consoles themselves, those that make games for multiple platforms will probably be lazy, before you flame, hear me out (I may be wrong, I just want to see what you lot think of this).



    For example: The Xbox is far more powerful than the PS2, however, the graphics in Burnout 3, a very popular game, are essentially the same on both consoles. Visually there is no dramatic difference, and there is certainly the opportunity for there to be.



    So from the information I've read, the Xbox 2 and PS3 will be very different beasts in terms of their system architecture. As far as I am aware, the PS2 and the Xbox are closer in terms of design than the Xbox 2 and PS3 will be. This could make porting more difficult, resulting in a larger portion of games reaching the market (that are made for both consoles) having a low level of optimisation, and therefore the potential power of these consoles will only be fully used by Sony and Microsoft themselves.



    So in terms of ratio, the graphics will be better in the next generation no doubt, but the standard could be worse than we expect.



    No?
  • Reply 117 of 134
    powerdocpowerdoc Posts: 8,123member
    Quote:

    Originally posted by mattyj

    Well the problem is, with the exception of the companies that make the consoles themselves, those that make games for multiple platforms will probably be lazy, before you flame, hear me out (I may be wrong, I just want to see what you lot think of this).



    For example: The Xbox is far more powerful than the PS2, however, the graphics in Burnout 3, a very popular game, are essentially the same on both consoles. Visually there is no dramatic difference, and there is certainly the opportunity for there to be.



    So from the information I've read, the Xbox 2 and PS3 will be very different beasts in terms of their system architecture. As far as I am aware, the PS2 and the Xbox are closer in terms of design than the Xbox 2 and PS3 will be. This could make porting more difficult, resulting in a larger portion of games reaching the market (that are made for both consoles) having a low level of optimisation, and therefore the potential power of these consoles will only be fully used by Sony and Microsoft themselves.



    So in terms of ratio, the graphics will be better in the next generation no doubt, but the standard could be worse than we expect.



    No?




    You are quite true, I was mostly refering to the programmers of the companies who sold the box.

    For the others, it will be an another story. It take some months or may be one year or two, before we will start to see games who will take full advantage of the new architecture.

    But compared to others software in the desktop world, it woul be much easier.
  • Reply 118 of 134
    snoopysnoopy Posts: 1,901member
    Regarding updating an application's code for a Cell processor, an interesting observation was made over at AN, on the Power PC thread. Essentially, the idea is that existing applications only need be updated for Tiger, to use the core services. The core services in Tiger then would utilize Cell's SPEs for a dramatic performance increase. In this way, all the new and difficult coding is done by Apple when developing Tiger and core services. Cell's PPE runs the majority of an application code, which is unchanged. It seems like a workable idea to me. Only applications that use SPEs for other tasks need to have the new code.
  • Reply 119 of 134
    Quote:

    Originally posted by snoopy

    Regarding updating an application's code for a Cell processor, an interesting observation was made over at AN, on the Power PC thread. Essentially, the idea is that existing applications only need be updated for Tiger, to use the core services. The core services in Tiger then would utilize Cell's SPEs for a dramatic performance increase. In this way, all the new and difficult coding is done by Apple when developing Tiger and core services. Cell's PPE runs the majority of an application code, which is unchanged. It seems like a workable idea to me. Only applications that use SPEs for other tasks need to have the new code.



    Very interesting. Obviously the Cell was designed with current programming trends kept in mind.
  • Reply 120 of 134
    Quote:

    Originally posted by mattyj

    Very interesting. Obviously the Cell was designed with current programming trends kept in mind.



    so the CoreXXX thingy from apple will each get it's own processor-core :-D
Sign In or Register to comment.