Re: [HACKERS] tracking context switches with perf record

2012-03-30 Thread Robert Haas
On Fri, Mar 30, 2012 at 10:07 PM, Jeff Janes wrote: > Which I've now done, and it actually goes the other way.  with -F50, > not only do almost all pgbench_accounts updates still lead to a clean, > but a good chunk of the updates to pgbench_tellers lead to a clean as > well.  If cleans are going t

Re: [HACKERS] tracking context switches with perf record

2012-03-30 Thread Jeff Janes
On Fri, Mar 30, 2012 at 9:27 AM, Robert Haas wrote: > > The more interesting waits, in my view anyway, are the ones that come > from LWLockAcquire, which account for nearly all of the semaphore > sleeps.  As you can see, XLogInsert accounts for over half of those, > and ProcArrayEndTransaction and

Re: [HACKERS] tracking context switches with perf record

2012-03-30 Thread Andres Freund
On Friday, March 30, 2012 06:27:36 PM Robert Haas wrote: > Probability=No, score=-3.2 required=5.0 tests=AWL,BAYES_00, > RCVD_IN_DNSWL_LOW autolearn=ham version=3.2.5 I've been playing around > with "perf record" on the IBM POWER7 machine, and it's pretty cool. One > of the things I don't like i

Re: [HACKERS] tracking context switches with perf record

2012-03-30 Thread Robert Haas
On Fri, Mar 30, 2012 at 1:30 PM, Greg Stark wrote: > On Fri, Mar 30, 2012 at 5:27 PM, Robert Haas wrote: >> If you expand that branch of the call tree, you find that all of them >> are coming eventually from secure_read; the server is waiting for a >> new query from the client.  This is, obviousl

Re: [HACKERS] tracking context switches with perf record

2012-03-30 Thread Tom Lane
Greg Stark writes: > On Fri, Mar 30, 2012 at 5:27 PM, Robert Haas wrote: >> If you expand that branch of the call tree, you find that all of them >> are coming eventually from secure_read; the server is waiting for a >> new query from the client.  This is, obviously, overhead we can't >> eliminat

Re: [HACKERS] tracking context switches with perf record

2012-03-30 Thread Greg Stark
On Fri, Mar 30, 2012 at 5:27 PM, Robert Haas wrote: > If you expand that branch of the call tree, you find that all of them > are coming eventually from secure_read; the server is waiting for a > new query from the client.  This is, obviously, overhead we can't > eliminate from this test; waiting

[HACKERS] tracking context switches with perf record

2012-03-30 Thread Robert Haas
I've been playing around with "perf record" on the IBM POWER7 machine, and it's pretty cool. One of the things I don't like is that some of the tools don't produce very nice text reports that you can cut and paste into an email - it's kind of a text-based GUI where you can zoom around and look at