[1594] in Release_7.7_team
Ultra 10's - something fishy?
daemon@ATHENA.MIT.EDU (Alex T Prengel)
Tue Jan 19 19:46:40 1999
To: miki@MIT.EDU, wdc@MIT.EDU, ghudson@MIT.EDU
Cc: release-team@MIT.EDU, facdev@MIT.EDU, nschmidt@MIT.EDU
Date: Tue, 19 Jan 1999 19:46:35 EST
From: Alex T Prengel <alexp@MIT.EDU>
Hello,
Hot on the heels of ccount's discovery of a numerical application that
completely locked up our Ultra 10 (requiring a power cycle reboot), I
have run into a different situation resulting in the same.
I've been trying to install a set of physics simulations for a faculty
member which I received in binary form- I believe they are SunOS
binaries but whatever they are, they run fine on everything except
Ultra 10s (ours, allegretto, anyway). It completely locks up the
machine. For a sample, you can try "add 6.730; ziman", but only do
this on an Ultra 10 that you can afford to risk having to power cycle
reboot. (there are a bunch of similar applications in
/mit/6.730/arch/sun4x_56/bin. This really shouldn't be happening- they
run fine on triad, an Ultra 5, and allegretto-temp, an Ultra 1, Sparc
4s and Sparc 5s.
I'm wondering if our Ultra 10s might have some incorrect system
software? I noticed that machtype -c returns Ultra-5_10 on both our
Ultra 5 (triad), and our Ultra 10 (allegretto). Is this right? I need
some way to distinguish these machine types as I want to put a
protection wrapper in the launch script, preventing a launch on Ultra
10s until this is figured out. How can I do that?
It might be that our machine is hosed and this is not a general problem-
can some brave soul try this on another Ultra 10?
Perplexedly yours,
Alex
