[Aces-support] Matlab fix, possible?

[Sai Ravela]

Hi admin:


Plese check
http://www.mathworks.com/support/solutions/data/1-OUVOD.html?solution=1-OUVO
D for the matlab fix.
People without SMP do not have to experience this problem we've been having
with matlab.

Have I been added to sudoers, so I can apply this fix out on a test node and
see if it works?

Thanks much,
Sai

Here are more details:

...I don't know if there is a work-around, but I'll explain the problem and
maybe someone out there can find one. The bug in some sense exists on
all platforms, but is manifest only on Redhat because Redhat will
occasionally allocate memory at addresses above 2^31, i.e., the
high-order bit of the address is on. If this happens at just the wrong
place, a piece of very old code does a signed comparison on this pointer
and causes the blow-up. It took a while to find this because Redhat
allocation is unpredictable and it was hard to get a protocol that would
cause this to happen often enough to figure out the problem.

So the work-around would be to cause Redhat to allocate all pointers
below 2^31. I am not enough of a Redhat expert to know if this is
possible.

sorry about the bug,
Mike Karr


-----Original Message-----
From: aces-support-bounces at mitgcm.org
[mailto:aces-support-bounces at mitgcm.org]On Behalf Of chris hill
Sent: Tuesday, December 07, 2004 8:11 AM
To: aces-support at mitgcm.org
Subject: [Aces-support] ACES office hours today


Hi All,

 There will be an ACES office hours session 11-12 today in 54-322 as
usual.
 Please feel free to come along and bring your problems to be solved.

see you later,

Chris

_______________________________________________
Aces-support mailing list
Aces-support at acesgrid.org
http://acesgrid.org/mailman/listinfo/aces-support