1. Welcome! Please take a few seconds to create your free account to post threads, make some friends, remove a few ads while surfing and much more. ClutchFans has been bringing fans together to talk Houston Sports since 1996. Join us!

New articles about IBM's plans for POWER6 and Cell

Discussion in 'BBS Hangout' started by RC Cola, Feb 8, 2006.

  1. RC Cola

    RC Cola Member

    Joined:
    Jun 11, 2002
    Messages:
    11,506
    Likes Received:
    1,349
    I mostly just wanted to post the Cell-IBM stuff, but since IBM has also recently revealed details about the POWER6 CPU, I'll go ahead and post that stuff too. Note that I don't really follow that news too much, so I don't really have much insight regarding the POWER6. Anyway...
    http://www.dailytech.com/article.aspx?newsid=670
    http://news.ft.com/cms/s/c41f2d94-973c-11da-82b7-0000779e2340.html
    And regarding Cell...
    http://news.yahoo.com/s/ap/20060208/ap_on_hi_te/ibm_cell_chip
    Another article about Cell:
    http://www.businessweek.com/technology/content/feb2006/tc20060208_094930.htm
    Yeah, I may have gone a little overboard with the number of articles. :eek:

    Regarding the IBM-Cell articles, I realize that IBM adopting Cell seems like a no-brainer, although it should be noted that there was some speculation as to whether or not IBM would adopt the tech for their own purposes; basically, they may have helped designed Cell, but they may have had no desire to actually use it themselves. Of course, that doesn't seem to be the case according to these articles. I also pointed out some of the early non-STI companies that are using or may be using Cell in the near-future (Mercury and Raytheon). So yeah, Cell has far surpassed the Emotion Engine (its predecessor so to speak) when it comes to uses outside of a gaming console. :)

    As for the POWER6, again, I'll leave that up to you guys to discuss if you'd like. It would be interesting to see some sort of Cell processor based on the POWER6 (maybe a modified POWER6 core?). Other than that though, I don't have much to offer there.

    FWIW, I believe IBM is supposed to show off a new Cell demo today. I might post about that later if any articles are written about it.
     
  2. Rocket G

    Rocket G Member

    Joined:
    Jun 23, 2004
    Messages:
    1,623
    Likes Received:
    8
    Two key things you didn't highlight:

    "But for the chip to gain wide acceptance, IBM needs to spur outside programmers to write software that takes advantage of Cell's prowess. That could prove a tougher task than usual because Cell is architected so differently from other chips."

    -This would make sense as the word is that these are expensive & complex to program for. I suspect that very, very few gaming companies will have the funding & the expertise needed to deliver the "eye-popping" graphics that are being hyped. Raw power isn't jack if you can't take adavantage of it. Same as the 360, I think it will be a while before we see any true next-gen power games for the PS3.

    Also:

    "One knock against Cell could be that it produces a relatively high amount of heat, which could keep it out of mobile devices."

    This should be a big STFU to the fanboys complaining about the 360's heat shedding & massive external power supply. The PS3 will have the same issues.
     
  3. RC Cola

    RC Cola Member

    Joined:
    Jun 11, 2002
    Messages:
    11,506
    Likes Received:
    1,349
    This actually wasn't intended to be a gaming related thread (or even something non-IBM related for that matter except for a few points), but I guess I'll go ahead and address those points.
    Actually, quite a few developers might have the "expertise" of programming for Cell if they developed a PS2 game, and it seems like even a developer like Bandai can something "next-gen" looking in a short time period. Programming for multiple cores/threads is something some developers have been doing for a while now; there's no doubting that Cell is definitely going to take some time and money in order to get the most out of it, but I bet game developers (specifically those that worked on the PS2) may be able to get the most out of the chip. From what I've read, it seems like Cell is in some ways easier and in some ways more complex than the EE. Considering what developers were able to get out of the EE (without any help from Sony really until late in the game), I think it will be interesting to see what they'll be able to get out of Cell. Of course, the same also applies to Xenon.

    For non-game software developers, I'm not sure if they would be as up to the task as game programmers would be, although to be honest, I'm not entirely sure what they're used to coding for. I'm sure Cell will be quite a challenge compared to any work they've done on a AMD, Intel, or POWER CPU (uh...non-Cell POWER CPU anyway), but I wonder how they write software for these large supercomputers. I believe a few CPUs out there (Niagra?) are like Cell in some regards, so some programmers used to supercomputers and/or other powerful devices might be used to parallelism to a degree. I don't know. FWIW, it did seem like Mercury got some Cell demos up and running pretty quickly and already it was a substantial improvement over what they had been using. If all these companies continue to support Cell, I imagine others will basically have to learn how to program for Cell (or something like it), whether they planned on doing so or not.

    Just saw this:
    http://news.com.com/IBM+refurbishing+blades+with+Cell,+PowerPC/2100-1010_3-6036943.html
    More people to try and "tame" Cell? :)

    I'm a little confused by that comment from the article. I was under the impression that Cell is pretty good (IIRC) when it comes to performance/watt (or whatever you want to call it). Right after that quote, the article says:
    It almost sounds like they're talking about either the Cell designed for the servers (dual-Cell IIRC) or the regular Cell CPU that they have designed (the "9 core" one at 3+GHz). If they're referring directly to those designs for Cell, then of course it won't be able to make it into mobile devices. But then again, there's no need for those designs for a mobile device; a lower clock combined with fewer SPEs should cut down on power and heat issues quite a bit, and the CPU would still be very powerful for a mobile device. I'm assuming that's what Turek meant by "future iterations."

    FWIW, the numbers out there for Cell and Xenon suggest that Cell@4GHz and with 8 SPEs would require only about half the power that Xenon does (like 40-45W to 80-85W). I don't think those are exactly accurate (slight variations in methods used to count power usage), but it is the only numbers out there right now. Since the Cell in the PS3 won't need as much power as that particular Cell design though (lower clock, 1 less SPE), there still might be at least a slight advantage for Cell in that regard. It might make sense too since I believe the hottest part of Cell is the PPE, which Xenon has three of basically; I think the SPEs are pretty good when it comes to power consumption and avoiding heat issues.

    Of course, I'm not saying the PS3 will avoid issues that the 360 underwent. Even if the Cell/Xenon numbers were accurate, the PS3 as a whole might be more similar to the 360 system as a whole when it comes to heat and power issues. I guess we'll have to wait and see.

    edit: Regarding the power-performance ratio:
    http://research.scea.com/research/html/CellGDC05/05.html

    http://ps3.ign.com/articles/615/615535p1.html
    I could probably find more, but that's some basic info there.
     
    #3 RC Cola, Feb 8, 2006
    Last edited: Feb 8, 2006

Share This Page