Messages in this thread | | | Date | Tue, 20 Dec 2011 10:48:26 -0500 | From | Vince Weaver <> | Subject | Re: [RFC 4/5] x86, perf: implements lwp-perf-integration (rc1) |
| |
On Tue, 20 Dec 2011, Ingo Molnar wrote: > Granted, LWP was mis-designed to quite a degree, those AMD chip > engineers should have talked to people who understand how modern > PMU abstractions are added to the OS kernel properly.
You do realize that LWP was probably in design 5+ years ago, at a time when most Linux kernel developers wanted nothing to do with perf counters, and thus anyone they did contact for help would have been from the since-rejected perfctr or perfmon2 camp.
Also, I'm sure Linux isn't the only Operating System that they had in mind when designing this functionality.
Running LWP through the kernel is a foolish idea. Does anyone have any numbers on what that would do to overhead?
perf_events creates huge overhead when doing self monitoring. For simple self-monintoring counter reads it is an *order of magnitude* worse than doing the same thing with perfctr. (see numbers here if you don't believe me: http://web.eecs.utk.edu/~vweaver1/projects/perf-events/benchmarks/rdtsc_overhead/ )
Vince
| |