_My_ home box is idle; it performs equally well with any scheduler, or with _no_ scheduler. _I_ needed the feature in order to try my hand at kernel hacking.
Others may need it if they have symmetric per-cpu threads and they wish to maximize throughput (possibly at the expense of latency). By binding a cpu to a thread, cpu switching overhead for that thread is avoided. An obvious example is distributed.net's crackers -- latency is not an issue; throughput is.
> better performances? If so really please run a fixed SMP scheduler:
>
> ftp://ftp.suse.com/pub/people/andrea/kernel-patches/my-2.2.13pre14/SMP-scheduler-2.2.11-E
>
> Please check if you can get better performances by binding your process to
> a fixed CPU than with my SMP scheduler heuristics.
>
Since the machine is doing no useful work other than compiling the kernel, I doubt I'd notice the difference. But I will take a look.
----------------------------------------------------------------
Get your free email from AltaVista at http://altavista.iname.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/