[rescue] SS600 hates my hypersparcs :(

[Dan Debertin]

I've been going back & forth with Mike Spooner about this, and we're
both a little stumped. I recently installed the latest Bridgepoint
PROM (2.14.3H, I believe) on my SS670, with a view to using a pair of
dual-CPU, 125MHZ/512K-cache HyperSPARC modules to replace my aging
SM520/SM521 pair.

The PROM upgrade went fine. With the HS modules, it passes POST --
which usually means your modules are going to work -- and boots
Solaris 2.5.1 as it should. However, the machine panics under any load
(i.e., compiling something), with a message about a suspect SIMM. The
SIMM in question is a different U-number every time, and is even on a
different expansion board every time, so my feeling is that it's lying
about that. I'm back to stability if I switch back to my old 50MHz
module set.

The puzzling thing is that there are reports of the smaller-cache
modules (256K on RT625) working on this machine. It's just these 512K
modules that have problems, probably because the cache controller is
different (RT626). But (using the MBUS guide as my reference) there is
evidence of other RT626-based modules working. Bewildering.

Anyway, as I said, I'm running Solaris 2.5.1, with 2.5.1_Recommended
patch-clusters current as of December or so. The modules in question
are known-good; they're running in the SS20 I'm typing this email
on. I've searched sunsolve for a relevant patch to no avail.

Anyone have similar modules working on this machine, or know of
something else I can try?

TIA,

Dan
-- 
Dan Debertin
airboss at nodewarrior.org
www.nodewarrior.org
3ffe:2900:1100:2::2

There is no magic.