Very interesting to see RH32 in the swap meet. I worked some on
verification boards for RH32 in '93 and again in '95.
Sounds like the board you have was made after I left in '95, since it
includes Orca FPGAs which I don't think came about until after that time.
We mostly used Actel FPGAs when I was there.
It was a fault-tolerant design which was supposed to detect a fault, then
back up and start re-executing code prior to the fault. The boards I worked
on had only 1 CPU chip, 2 MMUs (for for instructions (IMMU) and 1 for
operands (OMMU)), and a connector for an SCU (I forget what that was ...
system control unit?) daughtercard, as well as an FPU (floating point unit)
daughtercard. I think it was designed to support multiprocessor configs,
but we only tested single-processor, as we were bringing it up for the 1st
time.
As I recall, RH32 was started around '88, supposed to complete around '91,
but was about 2 years late in '93 when I started working there. I think it
was finally announced in '98:
http://www.findarticles.com/cf_dls/m0EKF/n2232_v44/21034857/p1/article.jhtml
There were also efforts by IBM to make silicon-on-saphire versions of their
RISC System 6000 CPUs which supposedly did not need the explicit hardware
fault tolerance used in RH32, since the process was inherently resistant to
radiation induced faults. I don't know if it was ever shown to work well.
Seems to be a lot of info on the 'net about compiling code for RH32, and
even references to it in source files. I guess they ported Linux to the
thing long after I had left. I recall TRW contracted Green Hills Software
to make tools for it.
-Rob Hulvey
Received on Thu May 13 2004 - 13:21:08 BST