[jan-jaap]

Just for fun, here's a side-by-side comparison of the IP4 vs the mystery board. There's roughly 2 years between them. The IP4 is largely built from standard 74xx series and some PAL chips, now notice what difference a couple of ASICs in the newer board makes for overall component density. And the IP4 doesn't even include ethernet, it needs a separate CMC/SGI VME board for that. Also, notice the number of changes to the IP4 in the form of blue wires everywhere.

I intend to spend some money to get better pictures: better lighting and some sort of rig to suspend the camera and take pictures "straight down" to avoid perspective distortion.


030-0011-001 IP4, 12.5MHz


030-8018-002 Mystery board

[Irinikus]

Very cool! This definitely has to be one of the rarer SGI artefacts out there!

[jan-jaap]

Since I have a Professional IRIS that's giving me nothing but issues I figured I had nothing to loose and gave this board a spin.

   

It took me a moment to figure out which port is the console port, but it does power up in the Professional IRIS chassis, it cycles the LED codes on the 7 segment display of the chassis, and boots to a PROM menu:

   
I guess what we're looking at here is a multi-head personal IRIS with zero graphics heads. Cool, but very useless 

[robespierre]

It looks like there is a tact switch on the board (at N104)? That's pretty strange.

[weblacky]

If I had to make some kind of guess based on what's on the board. I would say it looks like some form of fiber channel arbitrated loop RAID controller.

Here's my thinking, even though it's not as popular SGI did use copper DB9 connectors for FCAL a lot before we all agreed on fiber optic media.

There's all those memory banks in it which would make total sense if it was a RAID controller versus even a normal drive controller.

There's chips from LSI on it which to me screams drive controller.


There's a couple big items that aren't LSI which to me screams management or brain chip that uses drive controllers.

It has a battery holder to save the settings one might ordinarily need with a RAID controller.

The two bottom ports that look like DB 25 and DB9 serial could actually be how you interface to the unit. Instead of writing all this custom management software they could've simply gave you a cable to hook to your system's native serial and told you to start a terminal session to it and to type the following commands to manage your RAID volume.

There is however one very glaring issues with my guess: there are only three DB9 port cluster together and FCAL would normally be in pairs, most common stuff would have two channels with a pair of ports on each channel using the a A&B channels available on the drives for multipath or alternate path connections.

Perhaps if somebody can get the spec for FCAL they might find if there's a voltage difference between that and serial they could be measured on one of the ports to see if it is in fact expecting an FCAL copper connection.

But without a view of different connectors or more components that's my guess.

[robespierre]

There's chips from LSI on it which to me screams drive controller.

LSI Logic was founded in 1981, a decade before anybody was making Fibre Channel HBAs. Their main business during the 1980s was mask gate arrays, which is why so many of their chips are in graphics workstations like SGIs or Suns. They all carry "L1Axxxx" numbers.

The other company of note in standard-cell ASICs was VLSI Technology.

[jan-jaap]

It looks like there is a tact switch on the board (at N104)? That's pretty strange.

It's a reset switch. It's unusual indeed. There are some other things that don't make perfect sense:

• The board was designed to work with the Professional IRIS backplane. It has a WD SCSI controller that maps to the same user-defined pins on the VME P3 connector as the IP4 does. It interfaces correctly with the control panel on the drive bay tower, the 7 segment LED works. The reset switch there works too. I haven't tried the DIP switches, but expect those work as well.
  
• But it does not use the ribbon cable to connect to the keyboard / serial ports on the bulkhead, like IP4. You've got the same ports you'd find on the E-module of a 4D/25: keyboard, 2x serial, parallel and AUI (without retaining clip), but they're internal to the system
  

I don't really know how the graphics were supposed to work. There's an installation guide for the V30/V35 (a 6U VME set) on Bitsavers which indicates a ribbon cable was used to bus P2 from the CPU to the graphics, perhaps a similar mechanism was used here. Even if I had a TG-V20 graphics board I'd be reluctant to just hook that up. They probably made a 9U VME version of that too.
   

[robespierre]

The empty pads beside the P2 connector also are only 60 pins, which is strange because the GIO connector on the IP6 has 64 pins.