From owner-freebsd-current Thu Sep 11 14:11:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id OAA18373 for current-outgoing; Thu, 11 Sep 1997 14:11:33 -0700 (PDT) Received: from usr03.primenet.com (tlambert@usr03.primenet.com [206.165.6.203]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id OAA18360 for ; Thu, 11 Sep 1997 14:11:24 -0700 (PDT) Received: (from tlambert@localhost) by usr03.primenet.com (8.8.5/8.8.5) id OAA10788; Thu, 11 Sep 1997 14:11:15 -0700 (MST) From: Terry Lambert Message-Id: <199709112111.OAA10788@usr03.primenet.com> Subject: Re: lousy disk perf. under cpu load (was IDE vs SCSI) To: mal@algonet.se (Mats Lofkvist) Date: Thu, 11 Sep 1997 21:11:14 +0000 (GMT) Cc: current@FreeBSD.ORG In-Reply-To: <199709111931.VAA01882@kairos> from "Mats Lofkvist" at Sep 11, 97 09:31:14 pm X-Mailer: ELM [version 2.4 PL23] Content-Type: text Sender: owner-freebsd-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > The interrupt sharing of the ncr and ahc controllers indeed seemed > to be the problem. This should not be the case. The only thing I can think of is that the difference in shared vs. unshared is really abysmal for shared bacuase of the "did you generate this interrupt?" that gets asked in the shared case. I *thought* it was asked in all cases, shared or unshared, so the difference you are seeing really surprises me. You might want to volunteer to test code for Stefan, since you can apparently repeat the problem reliably. 8-(. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.