"MS takes sweet time with Xbox2" -Nvidia

I think visuals coming from all three consoles next gen will be so similar that noone in the right mind will bother being obsessed over technical differences.

* PS3 can put out 15000 realtime lights on the scene! Xbox 2 can do 17000!! Holy crap it pisses all over POS3! :LOL: *
 
marconelly! said:
I think visuals coming from all three consoles next gen will be so similar that noone in the right mind will bother being obsessed over technical differences.

* PS3 can put out 15000 realtime lights on the scene! Xbox 2 can do 17000!! Holy crap it pisses all over POS3! :LOL: *

Haha nintendo can d0 20000 and cook my breakfeast :)
 
Xbox2 won't have a CPU that will even touch what Cell will be able to do, so they will have to rely on the GPU like they did for Xbox.

Gee this CPU talk sure sounds a lot like "RISC will own x86". When did that happen? The best RISC CPUs have trouble besting an x86 processor, especially in the price/performance ratio --very important for consoles. I love how people go on about specs, but fail to realise how much money and R&D goes into x86 and what that all comes down to. The ISA is one of the worst but it's implementations are some of the highest performers and that's without charging a few k per chip.

I'll believe Cell being the greatest thing since sliced bread when I see it.

Yes, it's a very interesting architecture, heck I thought something like this would have been out way sooner, during the tech boom in the 90s actually. All because of things like your entire house (appliances et al) would have small computers which would talk to this big super server in a closet tucked away somewhere which would coordinate everything. I figured all of these would require expert systems and lots of small fast CPUs and CPUs composed of small fast CPUs.

Oh, IBM's process technology is only worthwhile if you pay through the nose, their cheaper stuff doesn't even come close to touching Intel.
 
Keep dreaming that Msoft will put a 1Tflop CPU into the Xbox2. Intel nor AMD will even have one out at the time.

PS3 will have the most powerfull CPU, deal with it. Just like with the PS2.

Although the GPU is going to be the thing to debate..
 
Although the GPU is going to be the thing to debate

A machine is complete until all it's parts are good enough.

They will not make a less than adequate GPU for a console machine. The GPU is almost the most important aspect.

Speng.
 
Sony's problem is that they think they can keep relying on CPU's for intense graphical work. The PS3 is just the PS2 all over again: some super powerful CPU to act as an "Emotion Engine" and a less powerful rasterizer to complete the pipeline (unless the PS3's architecture is radically changing from the PS2). Why not a dedicated GPU to handle all of the graphical computations and a super powerful CPU to take on everything else?

M$ and Nintendo don't need to rely on 1Tflop CPU's to handle intense graphics work. They'll just use insanely powerful GPU's that'll eat the PS3's CPU for lunch.
 
Keep dreaming that Msoft will put a 1Tflop CPU into the Xbox2. Intel nor AMD will even have one out at the time.

PS3 will have the most powerfull CPU, deal with it. Just like with the PS2.

Although the GPU is going to be the thing to debate..

Wrong, the MPU used in the PS2 is not in anyway categorically better than the CPUs in the Gamecube or Xbox.
 
I don´t see Nintendo doing anything competitive if they intend to launch at PS3´s launch frame, honestly. And DeathKnight, there hasn´t been a single word about PS3´s graphics rasterizer, or whatever handles the graphics. All we know is that from the patent is that it will have about 256 GFLOPS of performance.

Sorry to sound like a fanboy, because probably this doesn´t even make any sense, but:

1 TFLOPS CPU + 256 GFLOPS GPU = 1.256 :LOL:
 
"hey will not make a less than adequate GPU for a console machine."

You didn't understand what I mean't. I didn't say they would make a crappy GPU, I said that the GPU will determine the IQ of the next gen consoles. Sorry if I wasn't clear.
 
Saem said:
Keep dreaming that Msoft will put a 1Tflop CPU into the Xbox2. Intel nor AMD will even have one out at the time.

PS3 will have the most powerfull CPU, deal with it. Just like with the PS2.

Although the GPU is going to be the thing to debate..

Wrong, the MPU used in the PS2 is not in anyway categorically better than the CPUs in the Gamecube or Xbox.

It is not categorically worse either though is it? PS2 was out a year or more before than GC and XBOX wasn't it?.. just asking - don't flame me anyone!

I even liked the SH-4 Processors but that's me --> weird
 
It is not categorically worse either though is it? PS2 was out a year or more before than GC and XBOX wasn't it?.. just asking - don't flame me anyone!

This is basically my point. It's not the best, nor is it the worse. Each CPU has it's own strengths. But the argument gets complicated once you bring in the workloads their supposed to address.
 
Saem said:
The best RISC CPUs have trouble besting an x86 processor, especially in the price/performance ratio --very important for consoles.

Only in the high performance general purpose processor bracket, in the embedded realm x86 is the one at a huge disadvantage. When you start putting multiple processors on a die, and you care neither about legacy support nor performance on serial tasks, the requirements for the processor cores are closer to those for embedded processors than for general purpose processors. It's all about MIPS/die-area and MIPS/Watt.

I love how people go on about specs, but fail to realise how much money and R&D goes into x86 and what that all comes down to.

Almost all of the money goes to making processors which can execute serial tasks as fast as possible. This is mutually exclusive with efficiency for parallel tasks, or energy efficiency for that matter.

Marco
 
DeathKnight said:
Sony's problem is that they think they can keep relying on CPU's for intense graphical work.

How is that a problem? You go on to make a similarly awkward comment:

Why not a dedicated GPU to handle all of the graphical computations and a super powerful CPU to take on everything else?

Why rely on a "GPU" and not a "CPU"? Your comment strikes me as really odd as your differentiating between this generic "CPU" and "GPU" terminology as if the parts Sony would use were PC-centric in design - which is hardly the case.

Perhaps you could explain to me: (a) Whats the diffrence between the Cell patent and, say, the Nv3x when it comes to the non-raterization / sampling functions? (b) How you can make this sort of comment as the PC IHVs are moving towards a unified shader model themselves. With this distinction disapearing (between Fragment and Vertex computation), why is it so wrong for Sony to throw significant and flexible processing resources into the pipeline?

As I see it, your staring at an architecture which is 3-4 years away (from PC of today) and can't understand why this would be based on whats currently in the PC field. But, I think once you see how the NV4x and R4xx architecture are designed and the trend they start - you'll become a bit more quiet. ;)
 
I guess if Microsoft wanted to trample all over the PS 3 spec's, they could adopt the early 3dfx strategy of multiple chips manufactured on a mature process. If Sony is going to have one large chip as the heart of the PS 3, certainly it will be a challenge for them to get really good yields.
 
Only in the high performance general purpose processor bracket, in the embedded realm x86 is the one at a huge disadvantage. When you start putting multiple processors on a die, and you care neither about legacy support nor performance on serial tasks, the requirements for the processor cores are closer to those for embedded processors than for general purpose processors. It's all about MIPS/die-area and MIPS/Watt.

Almost all of the money goes to making processors which can execute serial tasks as fast as possible. This is mutually exclusive with efficiency for parallel tasks, or energy efficiency for that matter.

Agreed on all points. The thing is any task can be done serially, but that's not the case with parallel execution.

But we're talking about two things here. One is the CPU and it's ability to execute instructions and the other is the work load. Cell will be able to work on many tasks at once, there will be times where one can parallelize tasks. The XBox CPU, assuming it's x86 --I don't see why not-- will at least one task and do it really fast --lets assume no Hyper Threading for now. The question is when it comes to serial tasks how well will cell do. If a task requires a set of results, then the speed at which that task executes will be the speed at which the last sub-task finishes. Yeah, it's all stating the obvious, sorry.

My point is that I think that the CPU in the PS3 will likely have too large a work load and will end up being roughly on par when it comes to what can be done game-wise. I have a hard time believing that one can push computing power so dramatically far ahead as some like to claim.

I'm sure if someone has large budgets --monetary and mental-- like x86 they could build an insanely powerful machine which would in terms of hardware performance metrics be categorically superior to x86, the question is who has this budget? The amount of development past and present in x86 is hard to compete against.
 
I dont see any computationally intensive task in game engines using timestep simulation (which is to say almost all) which cant be parallelized.
 
If Sony can get Nvidia to do part of the Graphics Synth 3 / Visualizer
(as is rumored & even hinted at by Jen-Hsun Huang) Sony might be able to avoid the extremely disappointing feature list and image quality (for its time) of the Graphics Synth 1 used in PS2, and perhaps offset any advantage that XBox2 would otherwise have with an ATI R500~R600 based GPU.
 
Sony's problem is that they think they can keep relying on CPU's for intense graphical work.

Well, Sony aren't relying on Intel CPU ;)

Why not a dedicated GPU to handle all of the graphical computations and a super powerful CPU to take on everything else?

Well you have two chips, if you put everything to GPU, the GPU will be large and CPU small. If you have the choice you may want the chips to be similar in size.
 
"(as is rumored & even hinted at by Jen-Hsun Huang)"

I heard about something like this, but can you please elaborate? Thanks.
 
Back
Top