Scott Quinn wrote:
I recently acquired a SGI IRIS Indigo R4k machine that
was
nonfunctional. After a bit of troubleshooting I tracked it down to the
PM2 (processor module daughtercard, R4400 + oscillator + 1MB cache),
which has visible damage to two of the cache RAM chips (smoke holes and
cracks). I am trying to trace back the likely sequence of events that
lead to this happening so it doesn't happen again. AFAIK in the R4400SC
the cache memory is attached directly to the R4400 with the exception
of the power leads, and the R4400 has all cache control logic
integrated. The PSU voltages have been checked and are within specs,
with no excessive ripple.
In my experience, chips do not blow up without an external cause that
drastically increases the current flowing through the chip, however I
also have zero experience with SRAM chips failing in a "spectacular"
manner. Is this possible/likely? The other option seems to be the R4400
dying and taking the cache with it. Any ideas on where to proceed from
here?
The PM2 module is electriacally identical, and mechanically close enough to the modules
used in the Indigo2, so if you can score a battered I2 with R4400SC150 module, you can
stick that into the Indigo and have the fastest possible configuration of that model.
Maybe removing the cache chips will result in a functional PC module, but I never tried
that.
They are very nice machines. Built like tanks. Even a defunct one serves as a nice example
of how workstations *should* be constructed.
,xtG
tsooJ