Messages in this thread | | | Date | Wed, 26 Jun 2013 09:49:04 -0500 | From | Nathan Zimmer <> | Subject | Re: [RFC 2/2] x86_64, mm: Reinsert the absent memory |
| |
On Wed, Jun 26, 2013 at 02:14:30PM +0200, Ingo Molnar wrote: > > * Nathan Zimmer <nzimmer@sgi.com> wrote: > > > perf seems to struggle with 512 cpus, but I did get some data. > > Btw., mind outlining in what way it struggles? > > Thanks, > > Ingo
System interactivity take a nose dive, including the ability to halt perf. The commands do get through but they take a while. Other processes tend to make signifigantly reduced progress. Also it tends to slow down all running processes. My guess it is the NMIs overwhelming the system but I have not found a good way to profile perf gone wild so it is only a guess.
The issue mitigate somewhat by restrictring the number of cpus perf is recording or by not using the -g option, but sometimes I do need data on the whole system.
Nate
| |