sched-domain cleanups, sched-2.6.5-rc2-mm2-A3
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 5 lines · Nick Piggin
I'm with Martin here, we are just about to merge all this sched-domains stuff. So we should at least wait until after that. And of course, *nothing* gets changed without at least one benchmark that shows it improves something. So far nobody has come up to the plate with that.
Hmm? I post numbers all the time.
Quoted 4 lines
There are other things, like java, ervers, etc that use threads. The point is that we have never had this before, and nobody (until now) has been asking for it. And there are as yet no convincing benchmarks that even show best case improvements. And
I don't have hard numbers anymore (Andrea, do you remember the details?), but I think we had actually better benchmark results in some java benchmarks with an aggressively balancing scheduler in 2.4-aa on non NUMA machines.
-Andi
Quoted 12 lines · Andi Kleen
Quoted 5 lines · Nick Piggin
I'm with Martin here, we are just about to merge all this sched-domains stuff. So we should at least wait until after that. And of course, *nothing* gets changed without at least one benchmark that shows it improves something. So far nobody has come up to the plate with that.
Hmm? I post numbers all the time.
Yeah I know - it just so happened that you got *worse* performance from Ingo's balance-on-clone than by simply decreasing the balancing interval just now :P
Although it is probably just because balance-on-clone is not tuned well.
I've said all along that it will probably be a good thing to do if it is implemented correctly, but 1. we need some good numbers, and 2. let's not stuff it in at the same time sched-domains goes in because it is a fundamentally different concept.