rcowh@dropbear.co.rmit.oz (Warwick Heath) (11/14/89)
We have a fully configured Multimax (20 APC's) with 64Meg memory. It is running UMAXV 2.2h so that we can fully utilise shared memory. (as opposed to UMAX4.2). When running large shared memory jobs ( >20-30 Megabytes shared) we have programs that are guaranteed to cause kernal panics within 2 or 3 minutes. We have tried lots of things, playing around with the tuneable parameters, e.g. shared mem limits, user mem limits, to no avail. We can get rid of the problem by deconfiguring half of the CPU boards, any configuration above 5 CPU boards will crash. This problem has been assigned an SPR number by Encore, however it would be nice to hear if anyone out there has experienced the same problem/ worked out a way around it. ta, warwick heath. ACSnet: rcowh@dropbear.co.rmit.oz CSNET: rcowh@dropbear.co.rmit.oz.au ARPA: rcowh%dropbear.co.rmit.oz.au@uunet.uu.net BITNET: rcowh%dropbear.co.rmit.oz.au@CSNET-RELAY UUCP: ...!uunet!munnari!dropbear.co.rmit.oz.au!rcowh Royal Melbourne Institute of Technology, P.O. Box 2476V, Latrobe St, Melbourne, 3000, Australia