> > > b) "gives good performance on 12 processor" implies to me as a consumer
> > > poor performance from 13 to 64 :) or maybe even 2-11...
> >
> > Supports != scales well. Different bottlenecks come into play on different
> > sizes. I don't know where the number 12 came from, but your change
> > excludes scalability problems at all. Dishonest. OTOH 'good performance'
> > says nothing, so... <shrug>
>
> 12 was from the original draft5 version, and my change was to remove it.
> Saying it can scale up to 64 processors is key (remember, we need people
Do we?
> to realize that this is competitor to NT Server, since NT Workstation
And since when does NT run on a platform that allows 64-way SMP?
Or 12-way, for that matter? IIRC 64-processor box was UltraSparc one.
> is limited to 2 processors). I'm not familiar enough with the SMP
> support to know if the kernel is what scales worse than the hardware
> at a certain point (12 procs it would seem), but the original phrasing
> seemed to imply that we had up to 12 working and up to 64 kind-of working.
>
> Any kernel is going to have scaling issues getting up to 64 procs, I
Sure. Just that some address them better than we do.
> just want to make sure this press release is as positive as possible.
You know, 'positive' is a strange word. BTW, I really hope that
I'll never see this, erm, piece of work printed. Would cause *severe*
embarrasment. Linux is *not* marketdroid-driven, so WTF do we imitate
them?
-
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/