[Tutor] Windows vs Linux processing speed.

Steven D'Aprano steve at pearwood.info
Sun Oct 16 06:13:50 CEST 2011


Steven D'Aprano wrote:

> Very few program's speed are greatly dependent on raw processor speed. 
> Processor speed is one of the great marketing gimmicks of all time. Of 
> course it has *some* effect, but the bottleneck is almost never the CPU, 
> and usually the speed of getting data and/or code out of RAM and onto 
> the CPU and from their into the core for the instructions to be 

Sigh. /s/their/there

> executed. CPU cache faults are really, really expensive, so the bigger 
> the pipeline into the core, the fewer the cache faults.

And double sigh. Obviously I wasn't paying too much attention to what I 
was writing. Obviously the number of cache faults is determined by the 
size of the cache, not the size of the pipeline.

Generally speaking, a processor with a fast core but a small cache will 
not perform as well as a processor with a slower core but a bigger 
cache. Within reason -- obviously it depends on the nature of the code 
being executed, some code doesn't benefit much from a processor cache.




-- 
Steven



More information about the Tutor mailing list