> > i can show you pure user-space workloads that will not scale past 2 CPUs
> > on any x86 SMP system.
>
> Can you point me/us to any references on the SMP scalability of various
> workloads and/or how to diagnose them?
The critical one on a PC is almost always memory bandwidth. Running a tight
loop on one CPU doing chained dereferences to the same cache colour (ie
continually cycling lines through cache) will heavily dent performance of
the whole machine. You really do have to try hard to get x86 apps on heavily
smp scaled boxes running from cache
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed Aug 23 2000 - 21:00:05 EST