Re: async-calls squid3/src comm.cc,1.81.4.16,1.81.4.17

From: Alex Rousskov <rousskov@dont-contact.us>
Date: Tue, 08 Jan 2008 23:24:47 -0700

On Mon, 2008-01-07 at 23:59 +0200, Tsantilas Christos wrote:

> Henrik Nordström wrote:
> >
> > Adding this reshedule adds a noticeable overhead an delay, plus makes
> > tracing of the code flow more difficult.
>
> The hope is that we are gaining in stability and less complex code
> in upper levels.

And after giving it a little more though, I now expect the performance
difference between the old code and the [optimized] new code (i.e., the
comm_close change discussed in this thread) to be negligible: The FD
closing is not really postponed until the next main loop cycle. It is
only postponed by a few extra function calls.

Initial AsyncCalls code may be a bit slower but once extra mallocs are
removed, its performance should be comparable.

The above is just an informed speculation, of course. We will run
benchmarks and real-world comparisons to estimate the difference.

Alex.
Received on Tue Jan 08 2008 - 23:25:05 MST

This archive was generated by hypermail pre-2.1.9 : Wed Jan 30 2008 - 12:00:09 MST