[MPlayer-users] pentium4 bench

[Vitaly Luban]

Arpi wrote:

> note that p4 2hgz doesn't seem to be 2x (or more) faster than 1ghz celeron-2.
> it's strange, as it has very fast ram (400mhz rambus compared to 133mhz sdram)
> and also the cpu has double clockrate.
> maybe the code should be optimized differently for p4?
>

Not only, there's a lot of other stuff which is not sped up proportionally
to CPU/FSB clock increase, like how fast are memory chips themselves?
I bet they are internally the same 40-50ns SDRAM which gives you
only marginal memory speed increase despite of FSB speedup. Also,
CPU pipeline length vs branch misprediction may increase the cost of
mispredicted branches eating up performance. All in all, you dont get
twice the performance by simple two fold CPU/FSB clock increase,
to get it, you have to increase the clock of all the components involved,
including I/O (PCI bus and cards on it, disks), memory chips, and so on.

Vitaly.