From owner-freebsd-questions Thu Jul 27 04:27:57 1995 Return-Path: questions-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.11/8.6.6) id EAA18933 for questions-outgoing; Thu, 27 Jul 1995 04:27:57 -0700 Received: from elbereth.blueberry.co.uk (surfs-up.demon.co.uk [158.152.128.94]) by freefall.cdrom.com (8.6.11/8.6.6) with ESMTP id EAA18925 for ; Thu, 27 Jul 1995 04:27:52 -0700 Received: (from nik@localhost) by elbereth.blueberry.co.uk (8.6.11/8.6.9) id LAA01078 for questions@FreeBSD.org; Thu, 27 Jul 1995 11:47:53 GMT From: Nik Clayton Message-Id: <199507271147.LAA01078@elbereth.blueberry.co.uk> Subject: Pentiums and cache problems To: questions@FreeBSD.org Date: Thu, 27 Jul 1995 11:47:52 +0000 () X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1929 Sender: questions-owner@FreeBSD.org Precedence: bulk How do, Can anyone provide technical information as to the nature of the problems concerning some motherboards and the processor cache? My Pentium (100Mhz) has two caches, one external, one internal. During my intial install of FreeBSD, they were both on. The install would randomly reboot at about halfway through. Turning off the external cache fixed this problem. However, with the internal cache on, I still get the occasional reboot, particularly if there is extensive disk activity. For example, running X6 with 6 xterms up and a copy of XEmacs, and a make of xview-lib running, switching to another workspace in ctwm causes a reboot. I surmise that this is because of the disk activity (Adaptec 2940 SCSI system). I've got 16Mb RAM and 64Mb swap, which should be sufficient (for the moment). The reboots don't seem to be because of a paniced kernel. I have crash dumps on, and none of the reboots have yet created anything that `savecore' could salvage. Obviously, the solution is to turn the internal cache off. But if I do this then I take quite a performance hit. My benchmark is compiling the kernel. My (fairly standard) kernel takes about 15 minutes from the start of `config' to the end of the final `make' with the internal cache on. If I turn the internal cache off it takes well over two hours to compile. I know that if I turn around to my motherboard supplier they'll say "It works find under DOS/Windows/OS2/Linux, prove it's a bug in our motherboard". So can anyone provide me with technical information that I can give the manufacturer and say "fix this"? Cheers, N -- --+=[ Nik Clayton System Administration, Blueberry Design Ltd, ]=+-- --+=[ nik@blueberry.co.uk 1/9 Chelsea Harbour Design Centre ]=+-- --+=[ root@blueberry.co.uk London, SW10 0XE. Tel: 0171 351 3313 ]=+-- "It's two o'clock in the morning . . . do you know where your stack pointer is?"