Maximum bandwidth for multicores

So here's the question: If P4 with 800FSB can handle (64X800)mbits will a multicore of the same FSB be able to handle 2 or 4 x(64X800)mbits since he has 2 or 4 processors ?

If this is not the case then y would anyone get ddr3 modules?

Ta.

For Intel CPUs, it is the FSB (through the NB) that determines the amount of data that gets to the core or cores. The number of cores doesn't affect that speed. So in the future as more and more cores are added, you may not be able to feed them fast enough. This is one among several reasons to put the memory control on the CPU die.

With the memory controller on the CPU die, such as AMD currently uses and Intel plans to do in the future, the communication between memory and the CPU is faster (less latency). Memory speed in this case is controlled by what is called "the memory divider" and the speed is relative to the CPU speed, not the FSB speed.

So according to my calculations the "maximum" bandwidth throught fsb(1600 o/c) could be like 1600X64/8=12.8mb/s, doesnt that null the use of any ram beyond pc2 8000 in intel based CPU?

In terms of "stock speeds," you're right. But you can overclock the FSB further to get faster memory performance.

However, in practice unless you are doing heavy scientific calculations, video editing or some other computationally intensive exercise you aren't going to see memory bandwidth as a limiting factor.