65nm Cell; PlayStation cost controls on the horizon

Discussion in 'Beyond3D News' started by Carl B, Feb 5, 2007.

  1. Carl B

    Carl B Friends call me xbd
    Moderator Legend

    Joined:
    Feb 20, 2005
    Messages:
    6,266
    Likes Received:
    63
    Last week during Sony's quarter-end earnings call, CFO Takao Yuhara provided new insight into upcoming cost-cutting measures for the PlayStation 3, including a move already underway towards 65nm production for Cell.

    Asked by an analyst what Sony was doing to help reduce PS3 manufacturing costs, Yuhara indicated that besides the move to 65nm, the Playstation would be receiving a component update designed to 'reduce the parts' inside of the console. Whether this was in reference to the anticipated removal of the EE+GS chip used to power backwards compatability is unknown, but would certainly make sense. Sony has been working feverishly to perfect their PS2 software emulation; removing the legacy hardware would reduce both chip expense and motherboard complexity. Other potential areas for chip consolidation would be the unification of separate smaller ICs - such as bluetooth and WiFi - into single chip solutions and a change to a more specialized southbridge. Suspected of being Toshiba's 'Super Companion Chip,' the present southbridge occupies a large die area while providing superfluous functionality not presently required by the PS3.

    In terms of Cell on 65nm, Yuhara said only that production is presently underway. He did not state explicitly whether chips were being sourced from East Fishkill or Sony's own Nagasaki (or both), but confirmed that die size had been reduced by 40%. PlayStations shipping with the new Cell revision probably won't hit shelves until after the European launch, however, as SCE will wish to work through present supplies before transitioning. Going forward, as primary silicon gets shrunk, and tertiary silicon is removed from the system, PS3 should enjoy a cascading cost-reduction effect. Not only will the cost of the chips themselves go down, but the associated motherboard, cooling solution, and power supply will be able to be 'lightened' as well, as power and thermal pressures subside.

    Yuhara indicated seperately that as of December, blue laser diode yields have improved greatly, and he expects 6 million PS3's shipped by the end of Sony's fiscal year (March 31st, 2007) to remain an achievable goal.
     
    #1 Carl B, Feb 5, 2007
    Last edited by a moderator: Feb 5, 2007
  2. suryad

    Veteran

    Joined:
    Aug 20, 2004
    Messages:
    2,479
    Likes Received:
    16
    Just in time for me to get one...and an HDTV...and some new games for the PS3 library!
     
  3. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    214
    Location:
    Uffda-land
    They're still claiming they can get to 6M by the end of March? I'm torn between asking for a stick of whatever he's smoking, or trying to figure out how that could happen. Could they have been assembling more PS3s than they've been shipping, for adding BR drives to them as they become available?
     
  4. Carl B

    Carl B Friends call me xbd
    Moderator Legend

    Joined:
    Feb 20, 2005
    Messages:
    6,266
    Likes Received:
    63
    Could be, but at 1 million a month (which they seem to be capable of manufacturing), I think 5 million would be the most we could expect.
     
  5. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    214
    Location:
    Uffda-land
    At one point they claimed to be able to do 1.2M/month if they really needed to.
     
  6. Rangers

    Legend

    Joined:
    Aug 4, 2006
    Messages:
    12,658
    Likes Received:
    1,470
    Sales should be the bottleneck now.

    Sell through if we assume a lackluster 200k a month in America, and 100k in Japan..where are they selling 700k additional per month? Granted they should get a bump from EU launch..

    Also, are we really sure they intend to remove EE+GS? MS has had a nightmare providing software BC. Unless Sony has an easier go because of owning the PS2 hardware or something.
     
  7. fellix

    Veteran

    Joined:
    Dec 4, 2004
    Messages:
    3,516
    Likes Received:
    442
    Location:
    Varna, Bulgaria
    I wonder, whether the 65nm Cell will enable all the SPEs active (and one to remain for OS prime priority, of course)?
     
  8. Geo

    Geo Mostly Harmless
    Legend

    Joined:
    Apr 22, 2002
    Messages:
    9,116
    Likes Received:
    214
    Location:
    Uffda-land

    I wouldn't think so. Not until the channel is flooded. Which for a world-wide CE device like this with a proven track record, I'm guessing is something north of 1M units (as many as 2M even?) itself at any given moment once it's available everywhere. . .
     
  9. rendezvous

    Regular

    Joined:
    Jun 12, 2002
    Messages:
    347
    Likes Received:
    12
    Location:
    Lund, Sweden
    Think about the situation that we would be in then. A couple of millions of PS3s with 7 SPEs and then the rest with 8. Deveolpers would still have to target the ones with 7 SPE. Enabling special effects or improved performance for the units with 8 SPE would make the early adopters and the unlucky that may get a "defective" unit with only 7 SPE enbled feel fooled.

    So do you think they will enable the 8th SPE?
     
  10. fellix

    Veteran

    Joined:
    Dec 4, 2004
    Messages:
    3,516
    Likes Received:
    442
    Location:
    Varna, Bulgaria
    Well, I see it as an performance premium here, depend on what will be affected -- physics, hi-order geometry, encoding/decoding, and possibly the PS2 soft emulator, if it ever come to use so much of the vector power of Cell. Some sort of quality "degradation" could be implemented instead, say less collision detect precision, longer load/decompression load-times, and overall more aggressive LOD system, if the dev's don't want to touch the speed performance side of the things. Overall said, a little more work to do in the already bloated of hard labor coding environment, for this platform. :grin:

    As for the early adopters, that or either way, they have always sucked at this point of involvement with the new technologies. :lol:

    The other thing I want to see in the new Cell is larger L2 cache (1MB), because it is evident that the current implementation of 512K is quite scarce for such a bandwidth beast.

    Curiously also, will the 65nm Xenon CPU in XB360 bump it's L2 speed from now half of the core clock to a full, and what will be the odds of this?
     
    #10 fellix, Feb 5, 2007
    Last edited by a moderator: Feb 6, 2007
  11. TheAlSpark

    TheAlSpark Moderator
    Moderator Legend

    Joined:
    Feb 29, 2004
    Messages:
    21,917
    Likes Received:
    7,829
    Location:
    ಠ_ಠ
    About as likely as the higher clocked PS2. ;)
     
  12. akscott

    Newcomer

    Joined:
    Jan 18, 2007
    Messages:
    2
    Likes Received:
    0
    i find i t extremely unlikely that either consoles cpus will have any performance improvements. process shrinks are a normal part of all consoles lifespans. no console i am aware of has had any major peformance changes mid lifespan. There are some minor exceptions. latter generation ps2s' EEs are clocked at 299mhz from 294(1.7% increase); late model xbox went from 8gig hdds to 10. any real performance increase will undoubtely cause game devs to want to use it excluding owners with older consoles. thats what pcs are for not consoles.

    furthermore die space is not the limiting factor, yields are. all spes are present just not all enabled. moving to a new process will also not help immediately with yields, infact it will probably be slightly lower at first until the process is refined further, especially as STI has not yet released any 65nm processors commercially that im aware of.

    the only way i see the spe being activated is for use when running linux desktop but being explicity ruled out for game code. but even then the yields would have to be so good that turning on the extra spe would not have a seriously impact.
     
  13. Carl B

    Carl B Friends call me xbd
    Moderator Legend

    Joined:
    Feb 20, 2005
    Messages:
    6,266
    Likes Received:
    63
    No, I don't think so. 7 SPEs will always be the maximum the developers can target, so no reason to not take advtantage of better yields by sticking to the defect-buffer model of disabling one SPE.

    Cell *will* be getting some improvements down the line, but not the version inside of the PS3. Added performance simply can't/won't factor into PS3 development, as the hardware profile has already been locked. Expect improved Cells in servers, workstations, supercomputers, and such...

    As for the XeCPU, likewise a non-upgrade situation.
     
  14. pelly

    Newcomer

    Joined:
    Sep 5, 2002
    Messages:
    159
    Likes Received:
    5
    Location:
    San Jose, CA
    In terms of the chances any console will see any performance improvements over the course of its lifespan, Carl B. is entirely right. The "key" to consoles is the fact that developers are working with a static install-base. Whether talking about a PS3 the day it launches or a PS3 version three years from now, developers know they have the same overall hardware platform to work with. That is, they have the same performance constraints today as they will have for that console tomorrow. Think of what would happen if consoles weren't a "closed" system...

    Say a few years from now, Sony opted to raise cache sizes, "unlock" an SPE, add memory, etc...Developers now would have to choose whether to fully take advantage of this new performance headroom and implement new features or add eye candy. On one hand, developers are always aiming to add more to the game to help differentiate it from the other games on the market. However, there is an enormous initial install-base of "older" Playstation 3 consoles that would not have all the performance enhancements the latest PS3's would have. As a result, the decision to implement new features in the game to use the new hardware would ultimately result in a lackluster gaming experience for the initial install-base. Performance could be choppy with the older hardware or certain features not be fully supported. In either case, can you imagine seeing a game look different or run differently on two PS3's which are side-by-side (just with different build dates)?

    For better or for worse, the performance constraints for a console are poured in concrete the day the first system ships.
     
    #14 pelly, Feb 6, 2007
    Last edited by a moderator: Feb 6, 2007
  15. fellix

    Veteran

    Joined:
    Dec 4, 2004
    Messages:
    3,516
    Likes Received:
    442
    Location:
    Varna, Bulgaria
    No wonder there is so much diverge in the general public about the PC, as a competitive game platform. :wink:
    The question is, whether the latest trend in console development targets more for a PC-like platforms (in terms of services and software at all), but keeping the locked-and-steady HW base, or is trying to outrun the PC itself, rendering it as a obsolete concept, somehow.
     
  16. pelly

    Newcomer

    Joined:
    Sep 5, 2002
    Messages:
    159
    Likes Received:
    5
    Location:
    San Jose, CA
    The key word to describe the whole Console vs. PC debate for gaming is "convergence". Yes, this ugly word has seemingly been repeated by analysts over and over for the last decade. However, first the first time ever the term is finally finding an appropriate application...

    In short, we have a static platform (consoles) fighting a dynamic platform (PC's) for total rule over the living room. For the sake of simplicity, we'll ignore the fact that this space is currently ruled by a host of separate and specialized components and instead avert our attention towards the two platforms in question and this notion of "convergence".

    The PC is a double-edged sword of sorts in that it has phenomenal strengths, though it has just as many weaknesses of the same magnitude. Given the fact that it is a dynamic platform, consumers have the ability to configure or customize the platform to their own specific applications. Here, the market is saturated with a seemingly endless selection of hardware and software to tackle whatever needs the consumer might have. Unfortunately, this strength is also a weakness as compatibility becomes a critical topic of discussion. With the PC enthusiast aside, not many consumers are aware what a driver is let alone be comfortable searching for the appropriate one. In similar fashion, not many people would be successful troubleshooting an issue where a new piece of hardware or software isn't mixing well with the PC without having to contact some support center or call up a more PC-literate friend. Make no mistake, when things are playing nice there is no substitute for having the ability to have the absolute fastest hardware on the planet every 6 months and be able to witness the latest games running at obscene resolutions with all the latest visual bells and whistles enabled. However, even this strength is at the mercy of overall cost of ownership and developers who are burdened with the decision to put much development into supporting this enthusiast hardware. Instead, developers and producers are often far more interested in focusing on the level of hardware the largest install-base has as that will likely generate the most revenue. Afterall, what developer wants to sell a game that will run extremely poor on all the but the latest and greatest enthusiast hardware when that select few represents less than 5% of the total market?

    In stark contrast, we have the console. Unlike the more traditional consoles such as Super Nintendo or even the original Playstation, the latest generation of consoles is brining far more than gaming capabilities to the table. In the case of the PS3 for example, you have a Linux-based PC with 60GB hard-drive offering wireless internet capabilities, multimedia playack, a solid gaming system, discrete Blu-Ray player, and more for roughly $599. Although hardly a number to disregard and certainly high by console standards, it is sinfully cheaper than the $1500 or more an enthusiast class PC would easily cost. However, the true beauty of the console lies in the fact that the static platform simply works. There is "never" a worry that a game won't run properly on the system because developers are designing it to work on that one specific hardware configuration. In similar fashion, peripherals also simply "work" because they are also only forced to be compatible with a single hardware configuration. Aside from the latest addition of firmware updates (which are simply downloaded and automatically installed), there are no lengthy procedures required by consumers to keep things operating as they should. Rather, everything is hidden from the consumer and done quickly and cleanly through a simple file download. Add to this the fact that nearly anyone can walk up to a console and be able to figure out how to use the system and you have yet another advantage for this platform. Unfortunately, the static nature of the console prevents it from even being "upgraded" during its lifecycle. This renders the console a dinosaur of sorts a few short years into the lifecycle when compared with the PC as that platform is always evolving and advancing.

    In the end, there does not appear to be a clear victor in either camp. Granted, it is now far more appropriate for people to start talking about "convergence" in the living room and begin looking at a platform which will handle all of our needs. However, given the current status of each platform it would seem as though we are not quite to a point where we can easily determine the ideal solution. Rather, enthusiasts will continue experimenting with each platform and acting as guinea pigs until one camp finds the magic recipe and gets it all right. At that point (per usual), the mainstream will take notice and jump on board...Until then, we get to keep hearing people talk about convergence with no apparent solution in sight...

    Makes us all happy to be enthusiasts, right? ;)
     
  17. 3dilettante

    Legend Alpha

    Joined:
    Sep 15, 2003
    Messages:
    8,416
    Likes Received:
    4,172
    Location:
    Well within 3d
    I'd say almost "never". Until recently, we'd "never" heard of a patch for a console game. The price of increased complexity has forced some of the worst traits of PC technology onto the console side, though at a markedly lower level.

    The other downside to a simple market is that you pay for it. Looking at the insanely high price for what you get when you want to buy the 360's comparatively miniscule hard drive, you can see the danger to being too locked into a single vendor.

    The abstracted internals and restricted upgradability aren't just a convenience to non-technical people, they are a license to print money if the vendor so decides.

    addendum:
    Then of course, there's the problem when a vendor decides that it just won't allow or doesn't do enough towards maintaining compatibility (perhaps forcing compatibility only for future high-cost product lines). The number of Sony's "HDTVs" and the PS3 resolution mistatch, for example, are examples where things don't necessarily always work.
     
    #17 3dilettante, Feb 6, 2007
    Last edited by a moderator: Feb 6, 2007
Loading...

Share This Page

  • About Us

    Beyond3D has been around for over a decade and prides itself on being the best place on the web for in-depth, technically-driven discussion and analysis of 3D graphics hardware. If you love pixels and transistors, you've come to the right place!

    Beyond3D is proudly published by GPU Tools Ltd.
Loading...