[extropy-chat] A slow down in Moore's law?
Alfio Puglisi
puglisi at arcetri.astro.it
Mon Oct 18 18:03:14 UTC 2004
On Mon, 18 Oct 2004, Kevin Freels wrote:
>Some additional checking, which I admit was light,
>revealed that the P4 1.7 Ghz processor was unveiled in 2001.
>This is somewhat more than 18 months for a two-fold increase in
>processing speed.
Actually it would worse than that, because processing speed doesn't scale
1:1 with clock speed.
>Is there a new processor out that I don;t know about,
> or was the P4 released later than that?
>Or os there something else that I am missing entirely?
Just that clockspeed is a poor measure of performance. AMD Athlon64 and
Opteron chips run around 2 Ghz, but they process data as fast or faster
than P4s. Intel's own Itanium, IBM Power5 and the rest of the server
chips run between 1 and 2 Ghz, but go much faster than any desktop-level
offer from AMD or Intel (well, apart from P4's integer performance which
is really high).
Cache architecture and dimensions, memory speed, bus design (for
multiprocessor system), simultaneous multithreading (known as
Hyperthreading in the P4 case, SMT for the others), all play a large
factor in determining a processor performance, while clock speed is just
another factor.
Alfio
More information about the extropy-chat
mailing list