The general rule is that the larger the cache is; the less often you
will have to go to main (slower) memory to obtain needed bytes. This
effectively increases the processor speed because you don't have to wait
additional clock cycles for the data to be presented to the processor.
Cache architecture and locality of referenced code (and data) also have
an effect on performance. No one architecture is best for all
applications and each design incorperates tradeoffs that tend to best
optimize a different memory reference patern.
If you want to explore this further, I'd try a hardware forum.
Phil Sherman
p175 wrote:
People,
Can someone please explain the benefits one might see in a processors
with larger L2/3 cache than say a standard server processor of 1mb.
Would 8mb L3 cache really make a huge deal to performance ?