Re: [GIT PULL 0/8] Annotation weekly ponies delivery

[Ingo Molnar <mingo@kernel.org>]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Sat, May 12, 2012 at 2:25 PM, Arnaldo Carvalho de Melo
> <acme@infradead.org> wrote:
> >
> > Didn't notice it, but then I'm not updating the kernel that frequently,
> > will check on this Sandy Bridge nobo (i7-2920XM).
> 
> I think SNB has problems with PEBS and cycles anyway. I have a 
> westmere machine that I use for profiling for that reason.

I have a pre-westmere Nehalem that I use most for perf testing - 
that one appears to be working well both with default cycles and 
with cycles:pp. Will test on a Westmere.

btw., Arnaldo, could we make cycles:pp the default event on perf 
record and perf top? We have the new AMD IBS code in perf/core 
queued up for v3.5, which is roughly equivalent to cycles:p on 
Intel CPUs, so if we standardize on cycles:pp we'll get improved 
skidless (and on Intel pre-SB CPUs, precise) profiling output by 
default.

While at it I'd also suggest increasing the default sampling 
frequency, from 1000 Hz per CPU to at least 4Khz auto-freq or so 
- this should work well all across the board I think. CPUs are 
getting faster and command/app run times are getting shorter, 
1Khz is a bit low IMO.

Thanks,

	Ingo