SGI: Hardware

So my 1GB memory kit arrived today, two 512MB dimms (holy cow are they huge!). As soon as I got home I prepped an old Linux laptop I have with mincom and a laplink cable and proceaded to install the dimm's into the first unit in the pile. typed power up and the first O300 came to life, awesome! After I started the system I switch to console mode by hitting ctrl_d and watch as the memory check reports an error. The error is "swapping bank 0 with bank 1". Rats do I have a bad dimm? So i open the system, swapped the memory modules and powered it back up. same message.
Further down the console boot log it says:
memory enabled = 512MB
memory disabled = 256MB

Thinking I might have been sent two mismatched dimm's I took them out and inspected them carefully, even comparing chip part #'s. Identical. Okay lets try the next system, power up, and what do you know, no memory errors and "memory enabled = 1024MB". I went back to the first unit and checked the dimm sockets for dirt, nothing clean as a whistle. Tried installing the memory again and wiggleing the diims to make sure they are firmly seated and still the same memory error messages.

Now here is where it gets interesting, The first unit I opened had the two screws securing the access lid. The other four units have their screws missing when the memory was pulled. That person knew the memory was missing and never bothered to open the unit. This leads me to believe that the memory controller had gone bad before the system was decommissioned. From what I know, nodes can be operated without memory when linked to a unit or units with memory, correct? They just ran the unit without memory.

I have not yet tested the other 3 systems but I will tomorrow when I have some more time. Looks like one system is a partial lemon . Any way to fix this problem or is this a bad logic board?

Before panicking try resetting the POD (resetall, enableall - do a search for "enableall" "POD" "SGI" and you'll get instructions) (provided that the O3x follows the same convention as the O2x).

Similar, I had a stick of o300 RAM that would be disabled every so often and some variation of "enable" worked. I don't recall the exact syntax. It maybe have even been enableall... from "5. Enter Command Monitor".. on the o300 menu.

R.

Perhaps not your issue here, but I've had the 030-1060-xxx DIMMS show up with SMT components knocked off the PCB. Haven't tried plugging them into anything yet...

The suggestions to clear the power on diagnostic (aka POD) logs are good ones.

To access POD mode, stop at the PROM command line (item 5 in the PROM menu list), and sequentially run the following commands from the command line:

• pod
• go cac
• clearalllogs
• initalllogs
• flush
• reset (the system will restart)

After the systen restarts, go back into the PROM monitor and execute:

• enableall
• update
• reset (the system will restart)

If the memory errors don't reappear, you're probably safe.

If they do reappear then I'd suggest trying that pair of DIMMs in the second set of memory slots - slots 1 and 3 normally hold the first pair, slots 2 and 4 the second pair. The error message you received, ""swapping bank 0 with bank 1", makes it sound like the system has disabled bank 0 and is expecting to find the memory in bank 1. It's a long shot that may not work, but if it does it beats replacing the logic board.

You sir are MY HERO!

I went from this:


To this!



Thank you for the clear and precise instructions.

You're welcome - I'm glad to see it had a happy ending.

It's a good idea to go through the enable/clear stuff every time you change the hardware. You can technically add hardware and the Origin will find it no problem, but if you substitute hardware or remove hardware there can be difficulties. The PROM/system controller "remembers" what it saw, and if things are different it can lock stuff out as failed or not correctly identify the new hardware.