BFS Scheduler Update Brings SMT Nice Support

[ParticleBoard]

Correct me if I'm wrong but...

Isn't the current default scheduler in the linux kernel (CFS) loosely written by CK? I seem to remember him writing it then a kernel dev taking his idea, rewriting it and re-releasing it with no credit which caused a big stink. In the end Con quit the whole kernel dev thing for a while.

For that reason alone if he wrote a new scheduler that he says is better for "less then 4096 cpus" I'm inclined to believe it.

[Zan Lynx]

Correct me if I'm wrong but...

Isn't the current default scheduler in the linux kernel (CFS) loosely written by CK? I seem to remember him writing it then a kernel dev taking his idea, rewriting it and re-releasing it with no credit which caused a big stink. In the end Con quit the whole kernel dev thing for a while.

For that reason alone if he wrote a new scheduler that he says is better for "less then 4096 cpus" I'm inclined to believe it.

That is not the way I recall it.

Yes, the first so-called "Fair" scheduler was written by CK. However, Ingo Moln?r took the idea of a fair scheduler and wrote his very own version of it using a more complicated model.

I think it would be fair to say that the two schedulers are based on each other like Minix and Linux or ZFS and BTRFS. In other words, some of the same ideas, implemented entirely differently.

[grayhat]

See here about Con's work (RSDL and SD schedulers): http://en.wikipedia.org/wiki/Con_Kolivas and here about his resignation: http://apcmag.com/why_i_quit_kernel_...on_kolivas.htm