My XBox 360 vs PS3 comparison (mostly technical)

[Final-Fan]

While the RSX can acess the XDRAM, but at a significant latency penalty. Since the crux of his argument was that the split RAM desing was superior to the 360 pooled design, I think this is a somewhat major oversight - especially given how large textures this gen are.

In the OP MikeB refers to a well-known diagram of the 360's design which shows that its entire 512MB of RAM is connected to the GPU, and from there to the CPU, at 10.8 up/10.8 down. Is this not in fact the case? And if it is the case, is this not a significant disadvantage the 360's CPU faces, and in fact a greater bottleneck for the ENTIRE amount of memory than is true of HALF the amount of memory available to the RSX and Cell? It's therefore true that the 360's GPU will be able to utilize the entire amount of memory very quickly, but its CPU will not be able to utilize ANY of that memory at comparable speed.

Another thing he overlooked was the hardware tesselator in the 360s GPU. One of the things brought up here was the supposed lack of bandwidth between the CPU and GPU in the 360s design. However, the hardware tesselator can take a lower poly model and tesselate it into a higher poly model, which saves heaps of vertex data shuffling from the CPU to the GPU. This is also relavent because you hear PS3 devs talk about using the SPUs to do geometry, yet the tesselator can do many of these things, in hardware on the Xenos.

Isn't this effectively putting the CPU's workload on the GPU? This is indeed a nifty trick, but it seems to be at the expense of putting extra work on the shoulders of the Xenos, and when the time comes that the GPU will be pushed to its limit just doing its ordinary duties (as will inevitably happen a few years down the road), the crutch will be pulled away, the technique will become useless/counterproductive, and you'll be facing the same bandwidth limitation. (Supposed bandwidth limitation? Either make your counterargument or concede the point please.)