Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On machines with lots of CPU cores, pgbench can start eating up a lot
> of system time. Investigation reveals that the problem is with
> I patched it to use random_r() - the patch is attached - and here are
> the (rather gratifying) results of that test:
> Since a client-limited benchmark isn't very interesting, I think this
> change makes sense. Thoughts? Objections?
Portability, or rather lack of it. What about using erand48, which we
already have a dependency on (and substitute code for)?
regards, tom lane
In response to
pgsql-hackers by date
|Next:||From: daveg||Date: 2011-07-29 22:35:04|
|Subject: Re: error: could not find pg_class tuple for index 2662|
|Previous:||From: Robert Haas||Date: 2011-07-29 21:00:25|
|Subject: pgbench internal contention|