Hajaz said:
you know the damage is done for sony when big sites like anandtech and ign are reporting x360 to be better
Ugh....anandtech said no such thing, and IGN is regurgitating MS PR. The latter has absolutely no credibility.
MightyHedgehog said:
They explain why they're stacking general purpose performance of the two against each other for the reason that they feel that game code isn't as reliant on FP math (on CPU) as Cell processor's extreme FP performance would lead someone to believe. That's a reasonable statement, since you'd be relying on FP for 3D, effects, and simulations...which the GPU would be repsonsible for in large part through shaders.
Unless you want some pretty crappy graphics, simulation is done on the CPU. And the point is, next-generation games ARE going to be simulation-heavy, or should be. You want them to be more realistic, right?
It's funny how they downplay floating point performance and then turn around and talk about dot products..
That said, the SPEs are more than just floating point work horses.
MightyHedgehog said:
Physics simulations would need that FP capability, but XCPU already provides plenty of that
Not as much as PS3's CPU though
When's enough? edit - For example, Epic is talking about Cell offering similar performance for physics as a AEGIA physx chip. Do you think MS's cores could keep up with that? Or the one of them that you could spend on physics?
MightyHedgehog said:
It sounds more balanced given what an average title would probably need on the CPU.
Games don't need balance. They need some things more than others. And obviously MS are going to try and spin the distribution of required power in their favour. There's lies, damn lies, and statistics.
MightyHedgehog said:
Where the memory bandwidth comparison sounds ridiculous, it doesn't sound too ridiculous in the actual writing of the article where it is explained that the extreme amount of bandwidth provided by ATI for the GPU main part to the daughter part (where the EDRAM+logic is held) is to all but eliminate the need to send that same data through the buses as much as you would do with the PS3 and its RSX. By keeping the amount of data flow low to RAM and doing all of your buffer work on the GPU side before kicking it out for rasterization after it hits mian RAM
Rasterisation happens before "framebuffer work" or is part of it. The thing he neglects to mention is that X360 would be in a world of pain if it didn't have this. PS3 has twice the main system memory bandwidth, and doesn't require somewhere to offload framebuffer bandwidth usage. If they want to count internal bandwidth on the GPU, shall we start counting internal bandwidth on Cell? Especially since we now know it can be used for framebuffer operations?
MightyHedgehog said:
All MS is doing with that set of graphs is to show off the same way Sony did at the conference. I mean, Sony spent more than half of their PC talking about the specs of the system and showing off demos and pictures detailing what those numbers are supposed to bring us as a result. The charts, themselves, on MS' bit are ridiculous...but no more so than what Sony did at their PC if you consider why they bothered to put them up for people to see.
Sony never made such silly comparisons.
MS isn't bringing anything new to the table, it's just bringing spin. They invented the concept of FUD, and they're putting it to use again here.
