SS20 Solaris 9 - Strange Issue

Vintage hardware at it's finest. Could use the help if you can send
some my way.

I have a Sparcstation 20 setup complete with 2 x SM71's, 448mb ram,
and an 8MB SX VSIMM.

I have loaded the May 02 release of Solaris 9 with the 9_recommended
patch cluster.

PROBLEM: With the SM71's, the machine locks up hard under heavy load
after about 15 min. Doesn't produce a crash dump. Must power cycle
to continue.

I have tried:

- Replaced both processors with spares - no change
- Replaced the motherboard - no change
- Replaced power supply - no change
- Replaced all the RAM - no change
- Verified cooling and even supplemented with a fan and the top off
just for testing. - no change


Now, if I replace the SUPERsparcs with 2 ROSS processors (2x200Mhz).
The issue goes away. I'd really rather use the SM71s as they kick the
crap out of the ROSS modules for general desktop usage. But I've
tried 2 pair already and that didn't help.

So any ideas about whats going on? Obviously some part of the system
is getting stressed, but other than the SX VSIMM, I've replaced
everything already so I'm at a loss.

Thanks

On Aug 23, 12:07*pm, Bart <xc68...@gmail.com> wrote:
> Vintage hardware at it's finest. *Could use the help if you can send
> some my way.
>
> I have a Sparcstation 20 setup complete with 2 x SM71's, 448mb ram,
> and an 8MB SX VSIMM.
>
> I have loaded the May 02 release of Solaris 9 with the 9_recommended
> patch cluster.
>
> PROBLEM: With the SM71's, the machine locks up hard under heavy load
> after about 15 min. *Doesn't produce a crash dump. *Must power cycle
> to continue.
>
> I have tried:
>
> - *Replaced both processors with spares - no change
> - *Replaced the motherboard - no change
> - *Replaced power supply - no change
> - *Replaced all the RAM - no change
> - *Verified cooling and even supplemented with a fan and the top off
> just for testing. - no change
>
> Now, if I replace the SUPERsparcs with 2 ROSS processors (2x200Mhz).
> The issue goes away. *I'd really rather use the SM71s as they kick the
> crap out of the ROSS modules for general desktop usage. *But I've
> tried 2 pair already and that didn't help.
>
> So any ideas about whats going on? *Obviously some part of the system
> is getting stressed, but other than the SX VSIMM, I've replaced
> everything already so I'm at a loss.
>
> Thanks

Forgot to mention that a single SM71 (tested both extensively) work
fine as well.

Some interesting progress to report here. I have had success in
stabilizing the system by pinning the mbus speed to 40mhz. I don't
really notice any difference in speed, but clearly there is some
loss.

I should point out that with more testing on my part, I can say that
it was stable with a 50mhz mbus as long as I didn't try to run any of
the more modern software like seamonkey, thunderbird, pidgin, etc...
I say "more modern" in the since that these programs absolutely stress
the system more than what was around in 1995. The amount of context
switching and load rises dramatically when using these types of
programs. I think another factor is that cg14 graphics are putting
more load on the cpu for things that would have normally been
accelerated by the SX subsystem. So more r/w to the memory controller
on top of a normal load.

still working on trying to track down the stressed component.



Bart wrote:
> On Aug 23, 12:07 pm, Bart <xc68...@gmail.com> wrote:
> > Vintage hardware at it's finest. Could use the help if you can send
> > some my way.
> >
> > I have a Sparcstation 20 setup complete with 2 x SM71's, 448mb ram,
> > and an 8MB SX VSIMM.
> >
> > I have loaded the May 02 release of Solaris 9 with the 9_recommended
> > patch cluster.
> >
> > PROBLEM: With the SM71's, the machine locks up hard under heavy load
> > after about 15 min. Doesn't produce a crash dump. Must power cycle
> > to continue.
> >
> > I have tried:
> >
> > - Replaced both processors with spares - no change
> > - Replaced the motherboard - no change
> > - Replaced power supply - no change
> > - Replaced all the RAM - no change
> > - Verified cooling and even supplemented with a fan and the top off
> > just for testing. - no change
> >
> > Now, if I replace the SUPERsparcs with 2 ROSS processors (2x200Mhz).
> > The issue goes away. I'd really rather use the SM71s as they kick the
> > crap out of the ROSS modules for general desktop usage. But I've
> > tried 2 pair already and that didn't help.
> >
> > So any ideas about whats going on? Obviously some part of the system
> > is getting stressed, but other than the SX VSIMM, I've replaced
> > everything already so I'm at a loss.
> >
> > Thanks
>
> Forgot to mention that a single SM71 (tested both extensively) work
> fine as well.