| bdbdbd said: @Mifely: Viral marketing? I didn't quite follow you all the way through, first you say that this console generation broke Moores law, then you say that it's bad that 360 has multiple processors, but good when PS3 has. When you are making your precious Cell to have more performance, you basically have 3 options to do it: 1. Increase the number of processors. 2. Increase the number of transistors on the dice. 3. Have it running on higher clockcycles. As great as you think the Cell is, it's just a processor just like every processor before it. Whether you call it Cell or PPC doesn't have anything to do with its heat production. If you add the number of processors, it's going to generate extra heat, just because it needs extra space, which causes more resistance per conductor and adds more latency. And the same goes for adding extra transistors. So, what to do? We slow down the clockspeed to prevent the extra heat (there goes the BC), or make smaller dices, so that we can fit more transistors and/or more prosessors for the same space, so that the heat production decreases and latencies doesn't come an issue. Just because someone decided to do a super-sized DSP, doesn't mean it's immune to laws of the physics. Same arguments that you had against 360 goes for PS3 too. @Squiliam: Saying that the Cell in PS3 is bad, isn't actually correct, when you take into consideration, that it was supposed to handle multiple tasks simultanously, into which it's nearly perfect. |
In part I agree with you, but my argument about the PS3 being able to scale more readily boils down to the Cell and the 360 core being fundamentally different types of parallel machines. The 360 uses 3 relative transitor-heavy cores to due its work, and generates a lot of heat. The Cell uses *much* simpler SPUs to do most of the work (at the expense of a lot of programming details that the extra logic on the 360 can help with), and overall, despite being much more capable when running code that works well in a parallel scheme, runs cooler than the 360.
The Cell is in a much better position to upscale, because it can afford to get a little hotter, with some process shrinking and more SPUs, whereas the 360 can't upscale as easily (this isn't to say "at all"). The reason Sony won't have trouble "upgrading" to a PS4 Cell is that its wacky programming requirements will remain the same -- the concept of localstore memory can stay the same, small processors with all their instructions and data right there, etc. All that logic dedicated to making a mondo-sized CPU (or 3) just isn't there, and PS3 apps have to be written to use the SPUs as intended, or be forced to run on the single, dual HW thread PPU. The PS3 has already headed down an upgradable architecture path, from a *software* perspective. The 360's software is going to be much more dependant on having those 3 3.2 GHz fully functional cores -- no 360 app will get by without them, plain and simple. Having to include that triple core processor into a new architecture, by 2011, is asking MS to drop a *lot* of money on something that might be worthless by 2013, whereas the PS3 can merely be scaled up for the next gen, since it already runs cooler.
I never said it was a freebie to upscale the PS3, or whatnot. Of course its a challenge -- I just think its less of one for Sony than it is for MS, and thus, given all things considered (esp. the success of the Wii), a new console generation by 2011 just doesn't seem feasible. Basically all I'm saying is that, because both machines are multi HW-thread architectures, both must retain their previous architures to upscale. I think the Cell is in a better position to upgrade than the 360 is. If every SPU had 1M of localstore memory, and that was the *only* upgrade, writing a parallel app on the PS3 would be piece of cake. That alone could make for a PS4. If you could also double the number of tiny SPUs (they really are very small, relative to the PPU and 360 cores), you would see a serious performance upgrade, and all developers would have to do, to upgrade their engines, would be to increase the number of SPU jobs they issue at once, and move even more stuff over to them, with the new, bigger localstores. Putting 6-8 big cores on the 360 processor is no joke, however, and would be very expensive, as I understand it. You could physically squeeze them onto the chip by 2011, no problem... dealing with the awesome heat that the (mostly unused, but still heat generating) extra logic bleeds out... not so easy.
Call it "viral marketing" if you like. I'm just calling it how I see it. If it bugs you that you can't think of a way for MS to upscale as easily, or my arguments just rub you that wrong way... sorry. I'm not a fan of the PS3, the 360, or the Wii, really. I just have taken note that the Cell is the processor of this current gen, that appears to be the most easily upscaled to the "next gen" -- unless you consider that the Wii can be easily upscaled into a HD platform, which of course, it can. I don't feel sad about it... sorry. MS pushed the 360 out the door to beat Sony and Nintendo to the draw, and its worked great for them thusfar. I just think Sony's plan goes forward to the next gen easier than MS's does, and that Nintendo has its work cut out, and is in the best position overall to upgrade.
If MS upgrades the 360, it'll be with a BD drive, and a MS-version of the Wiimote, along with a smaller box, and built in HDD. Plays all 360 games, and tries to challenge the Wii (because the 360s processor will be really cheap by then, presumably).
Microsoft ThWii-Sixty, by 2011. That I can totally see. Heck Nintendo might come out with a 2+ GHz SuperWii-HD to compete. But they won't really be the "next gen". Does that count? Maybe it does.
