Many game developers (not pointing figures) have used a percentage of a console's performance to try to hype up the performance of their game. Some developers have stated numbers near or even at 100%, which to me sounds pretty far-fetched. One reason: although current-gen consoles have amazing CPUs, the memory can't keep it fed. A crazy expensive CPU is pretty useless when the memory can only feed it data a fraction of the time. What I wonder is, when will console manufacturers focus more on memory bandwith and try to reduce the memory bottleneck? What technology could help improve memory bandwith?

bandwidth?

Are you referring to the Red Faction developer's comments on the 360?

It's all about how effeciently you use those resources.

DDR3 @ 2000MHz with three channels is plenty.

No, it's not.

22.4GBS memory bandwidth sounds like a lot, but it's not enough to do complex graphical operations like HDR, alpha-blending, and anti-aliasing at 30/60 fps.

The 360 does most of that in the 10MB frame buffer, and it has a 256GBS of memory bandwidth between the frame buffer and GPU.

there isnt anything as plenty, but the 360 uses tile rendering(like the DC did) so that saves it the need to process a lot of extra unneccesary polygons

http://en.wikipedia.org/wiki/Tiled_rendering

Bandwidth? Any embedded memory are bandwidth behemoths, they're like 5120-bit...
XDR2 RAM at 8 GHz looks pretty nasty also.

Actually massive memory bandwidth is kinda only really needed on the PC because they have to target such rediculously high resolution/AA levels for their high end cards. Other than that the memory bandwidth is pretty much no more and no less than what is required on the current HD consoles.

The Cell setup is excellent for throughput.

- First step of course is to not share main Ram between GPU and CPU like on the 360, separate buses maximise potential (less latency).
- Low latency main RAM, like XDR is crucial for getting the most out of a 8 processor chip like the PS3 Cell chip.
- But most importantly IMO are the SPE's dedicate memory. They function as real system memory, but operates as fast as cache memory. This together with the huge EIB ring bandwidth is what allows the Cell to reach a near linear performance increases when tapping more SPEs.