scooby_dooby said:
This is a classic example of why people here will always continue arguing something that simply never will be established. When comparing processors, you usually compare them based on their potential as this is what developers will strive to max out over time. The problem with comparing "real world performance" is that everyone has their own definition of it. To some, it's an example of their favorite game A on platform B from developer C looking better than game X on platform Y from developer Z. It just doesn't work that way and by comparing it like that, you're limiting the potential of the systems performance on the ability of that particular developer. By doing that, you missing out on valid factors like time, money, budget, stress, inefficiencies, libraries, team sizes... how do you quantify these factors that play a significant factor in every single game? Answer: You can't. What's left is a flawed conclusion, one that usually is only ment to suit the arguing persons agenda.
IMO the only way you'll ever get some reasonable and valid conclusion when comparing different things is if you messure them under the same conditions: Take a relative simple task and see which one results in better performance. Even then, it's still an apples / oranges comparison, because you might be ignoring that each processor / architecture tries to solve its problem differently from the other. In the case of PS2 / Xbox for example, we have a system [Xbox] that allowed developers to extract easy performance while doing complex pixel operations, while on the other [PS2] developers have a very flexible yet complicated system that's optimal usage lies in different areas. When both are used efficiently and to the systems strength, you get very different results: I'd say a game like ZOE2 shows what the PS2 is very good at: Geometry, particles while low on textures. Xbox's advantage in my eyes lie primarely in its pixel operations (bump mapping) and texturing... something the PS2 isn't good at. After 5 years of this generation, it's still arguable which of these two systems is "more powerful" - but perhaps only because it has ever been an apples / oranges comparison from the beginning. In programmable performance, I'd give the edge to PS2 among other advantage like more usuable fillrate - on Xbox, the advantages clearly goes into other aspects of its hardware. How do you quantify which is "better"? In the end, it simply boils down to personal preference.
Fast forwarding back to the PS3 / Xbox360 situation, I can't help but feel some of you are really looking at it from the wrong angle. The CELLs potential is its floating point performance - something that will give it a lot of potential for developers to use. This is an undeniable fact - how far it will be utilized by developers is a completely different discussion, one that even in 5 years from now won't give you cut and dry answers. There will be developers extracting better performance than others - and none of them is a definite benchmark of a systems performance because they're influenced by many factors (cost, teamsize, effort...) as well.
As to which of the two systems will be "more powerful" (what a silly term)? Does it really matter when the results are different? I expect each exclusive developers to make brilliant use of the hardware each going for different looks based on the system's potentials while multiplatform targeted games will rarely be of any benchmark since they will be programmed to suit both systems at the same time. If we want to be geeks though and compare processors, you have to see the processors for what they are and what the approach they require. As such, CELLs potentials are undeniable (as is XeCPU and any other unit of each of the systems).
It's plain and simple to me and should be to you.
