Received: from localhost (unknown [200.46.204.183]) by developer.postgresql.org (Postfix) with ESMTP id 4AF3D2E0041 for ; Wed, 11 Jun 2008 22:56:40 -0300 (ADT) Received: from developer.postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.183]) (amavisd-maia, port 10024) with ESMTP id 95729-03 for ; Wed, 11 Jun 2008 22:56:11 -0300 (ADT) X-Greylist: delayed 00:15:52.612578 by SQLgrey-1.7.6 Received: from p02c11o145.mxlogic.net (p02c11o145.mxlogic.net [208.65.144.78]) by developer.postgresql.org (Postfix) with ESMTP id D888B2E0040 for ; Wed, 11 Jun 2008 22:56:15 -0300 (ADT) Received: from unknown [198.6.208.103] (EHLO p02c11o145.mxlogic.net) by p02c11o145.mxlogic.net (mxl_mta-5.5.0-3) with ESMTP id f3280584.3196418992.309139.00-560.p02c11o145.mxlogic.net (envelope-from ); Wed, 11 Jun 2008 19:56:15 -0600 (MDT) Received: from unknown [198.6.208.103] (EHLO mail.sonalysts.com) by p02c11o145.mxlogic.net (mxl_mta-5.5.0-3) with ESMTP id 68e70584.2682416048.308588.00-006.p02c11o145.mxlogic.net (envelope-from ); Wed, 11 Jun 2008 19:40:22 -0600 (MDT) Received: from 206.53.65.108 ([206.53.65.108]) by mail.sonalysts.com for pgsql-performance@postgresql.org; Wed, 11 Jun 2008 21:40:20 -0400 To: pgsql-performance@postgresql.org From: "Josh Rovero" Subject: 8.3.1 vs 8.2.X on HP-UX PA-RISC 11.11/11.23 Message-ID: <20080612014020.431ffdde@mail.sonalysts.com> Date: Wed, 11 Jun 2008 21:40:20 -0400 X-User-Agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9) Gecko/2008052912 Firefox/3.0 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Spam: [F=0.2000000000; S=0.200(2008052701)] X-MAIL-FROM: X-SOURCE-IP: [198.6.208.103] X-Virus-Scanned: Maia Mailguard 1.0.1 X-Spam-Status: No, hits=2.599 tagged_above=0 required=5 tests=RCVD_NUMERIC_HELO=2.599 X-Spam-Level: ** X-Archive-Number: 200806/49 X-Sequence-Number: 30399 We run GCC-compiled postgresql on a number of HP-UX and Linux boxes. Our measurements to date show 8.3.1 performance to be about 30% *worse* than 8.2 on HP-UX for the same "drink the firehose" insert/update/delete benchmarks. Linux performance is fine. Tweaking the new 8.3.1 synchronous_commit and bg writer delays that *should* speed things up actually makes them a bit worse, again only on HP-UX PA-RISK 11.11 and 11.23. Right now it's 32 bit, both for 8.2 and 8.3. Any hints? P. J. Rovero