Received: from localhost (maia-3.hub.org [200.46.204.184]) by postgresql.org (Postfix) with ESMTP id 26F029FBF1C for ; Tue, 8 May 2007 19:31:14 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.184]) (amavisd-maia, port 10024) with ESMTP id 21382-02 for ; Tue, 8 May 2007 19:31:06 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.4 Received: from mail.net-domino.com (www.net-domino.com [74.208.11.96]) by postgresql.org (Postfix) with ESMTP id 01A1E9FB632 for ; Tue, 8 May 2007 19:31:09 -0300 (ADT) Received: from [127.0.0.1] (u15245314.onlinehome-server.com [74.208.11.96]) by mail.net-domino.com (8.13.7/8.13.6) with ESMTP id l48MUvWP024775 for ; Tue, 8 May 2007 18:31:02 -0400 Message-ID: <4640FA6E.4050003@mochima.com> Date: Tue, 08 May 2007 18:32:14 -0400 From: Carlos Moreno User-Agent: Thunderbird 2.0.0.0 (X11/20070326) MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Throttling PostgreSQL's CPU usage References: <20070508205120.GA7983@uio.no> In-Reply-To: <20070508205120.GA7983@uio.no> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Added-Header: Sample_milter accepted message X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200705/149 X-Sequence-Number: 24539 Steinar H. Gunderson wrote: > Or use a dual-core system. :-) Am I missing something?? There is just *one* instance of this idea in, what, four replies?? I find it so obvious, and so obviously the only solution that has any hope to work, that it makes me think I'm missing something ... Is it that multiple PostgreSQL processes will end up monopolizing as many CPU cores as you give it? (ok, that would suck, for sure :-)) If there is a way to guarantee (or at least to encourage) that PG will not use more than one, or even two cores, then a quad-core machine looks like a promising solution... One thing feels kind of certain to me: the kind of system that the OP describes has a most-demanding need for *extremely high* CPU power --- multi-core, or multi-CPU, would seem the better solution anyway, since it promotes responsiveness more than raw CPU power. Carlos --