Re: Times do not add up in 10046 trace file

From: <mccmx_at_hotmail.com>
Date: 13 Oct 2006 00:53:56 -0700
Message-ID: <1160726036.561860.124030@k70g2000cwa.googlegroups.com>

> Did you ever end up running it through orarsp? I wrote a trace parser
> a while back that printed statements in execution order, and the
> elapsed time between each. It has proven to be very useful to me for
> this kind of thing, but wouldn't be of use to you in this specific
> case. I would be curious to see what you get from something like
> orarsp. I can say that I have always seen aberrations in the timings,
> but never anything as large as the disparity you are seeing.
>

Yeah just ran it through yesterday. I found that orasrp was much better at pinpointing the root of a performance problem than tkprof. i.e. tkprof wasnt able to combine 'similar' statements (not using bind variables) but orasrp was.

However it didn't shed any light on the difference between elapsed and cpu time overall.

Matt Received on Fri Oct 13 2006 - 02:53:56 CDT