[weblacky]

Well to keep this topic going, maybe another valid question is, does anyone know what IC the PROM’s flashed image is stored in?

I assume it’s on an IO9? Could one desoldered the working IC, place in a socket on the IO9, dump contents of desoldered IC using some kind of programmer, then if anything goes wrong- restore IC contents from previous backup and keep trying?

I know the older stations had a single chip for the PROM. While I’d imagine the IO9 to be more integrated, if you backup your own IC and restore it, I’d imagine everything else on the board would be in sync?

Any thoughts?

[indy99]

Well to keep this topic going, maybe another valid question is, does anyone know what IC the PROM’s flashed image is stored in?

I assume it’s on an IO9?

Thought you were working on a Fuel? If you'd like to take a reasonably close up look at an IO9 look here's the best image I can offer without pulling a system out of the rack.

[weblacky]

Sorry,
I don’t own an Fuel (all future planning, I thought they used an IO9 design like the Tezro, which I do have). I assume the socketed PLCC chip on the IO9 is currently the Flash location.

I guess they don’t, I can’t find a closeup picture of the Fuel mainboard on Google images. So I guessed it used a similar PCB backplane.

I see from the manual it doesn’t (which means it’s on the mainboard). Again, no pictures.

But since both have an L1, I figured maybe they had a common implementation. So I guess I’ll go with the same answer (likely a socketed PLCC).

I’ve not found any info on manual PROM dumping new stations (only older ones). I was hoping that someone had already done these things. Perhaps the fact they haven’t means the flash region isn’t accessible or are spread over several ICs?

[hamei]

here's the best image I can offer without pulling a system out of the rack.

Let me assist, since I seem to have parts all over the place ...

   

   

[weblacky]

Thanks,
I’m suspicious of the ICs with the yellow stickers on them. The small one looks like a CMOS chip (just visually). But being covered I can’t tell, the printing means nothing to google (so likely SGI date/version codes).

At least that narrows down ICs. Both labeled ICs look like the only socketed ICs as well. So good bet that’s for a reason.

Ok, back to Fuel then, anyone got a source for high-res mainboard images?

Thanks!

[hamei]

I guess they don’t, I can’t find a closeup picture of the Fuel mainboard on Google images.

That's because google sucks balls and we should not be trafficking with the devil. But have no fear, underdog is in the neighborhood ...

   

and an 800 mhz cpu for good luck ...

   

Good thing I happened to find these again this morning, when I was looking for something else. The pimm connects via the white compression connector thingy at the top of the mainboard photo.

But forget the messing around nonsense, just set the speed the correct way.

Whoa, cpu came out okay for a camera photo. Sorry about the mainboard, flash :-(

Now if soeone would just tell me why the damn O350 insists on booting to the console .....

[indy99]

I’ve not found any info on manual PROM dumping new stations (only older ones). I was hoping that someone had already done these things. Perhaps the fact they haven’t means the flash region isn’t accessible or are spread over several ICs?

The L1 command "eeprom" outputs a stream of hexadecimal data that might or might not offer what you're looking for.

Documentation I have access to is limited to the l1 command "hlp eeprom"; which returns

Code:

001c01-L1>hlp eeprom
eeprom
           show brick eeprom data

eeprom <exp> <exp> <exp>
           show brick eeprom data at <eeprom> <offset> <length>

eeprom Fuel write default
           write standard Fuel EEPROM sata to MAC EEPROM

An L1 eeprom run on my Tezro returns data in hex format for a number of hardware system, including but not limited to the node, DIMM(s), Odyssey, and Snowball (DM3). None of my SGI systems are connected to the internet, so forgive me for not manually re-entering a couple of hundred lines of code.

In contrast to the IC rich Fuel system board, here's one of a nodeless O350 for contrast:

[hamei]

In contrast to the IC rich Fuel system board, here's one of a nodeless O350 for contrast:

Can you numalink to that with no nodeboard ? What happens if you stick a graphics card in there, will a numalink cable on the XIO slot give you dual head without needing processors and memory ?

From what I can tell, it seemed like you'd need at least an mpx brick to get two graphics heads but ... since you have that handy and all ...

[indy99]

From what I can tell, it seemed like you'd need at least an mpx brick to get two graphics heads but ... since you have that handy and all ...

Not quite so handy; isn't/never was my box. (-; I grabbed the photos so I'd have some shots of the somewhat obscure O350 PCI expansion module. Interesting set up. No node board/memory, no IO9, not even any serial or PS/2 ports. Considering what SGI charged for processor nodes and the dearth of available PCI slots in an O350, it must have made sense at some point in the game.

Can you numalink to that with no nodeboard ? What happens if you stick a graphics card in there, will a numalink cable on the XIO slot give you dual head without needing processors and memory ?

NumaLink yes, that much is documented in the O350 UM. Since the PCI slots are intended for use, and the XIO slot used by graphics is on the same riser, it *ought to* work. If you take into count that the even less often seen O350-based Onyx4 had a nodeless box with not much more than the dual slot AGP riser attached to what looks like the same system board, I'd think the odds are at least favorable.

Be another of those not as SGI intended deals we all seem to be so fond of.

Speaking of AGP risers, I have one boxed up around here somewhere. Pretty sure this one came came out of a gutted Prism rack system, which is why I never had the balls to stick it in a perfectly good running O350.

[kaigan]

I can confirm that the system board between the Onyx4 G2 brick (the one without a node board) is the same as the one in an O350 expansion unit. The O350-based stuff all uses the same 2U Interface board, 003-1809-00x. Also, the only difference between the Onyx4 G2 and the G2N is literally the node board. They're otherwise identical.

I'd also be curious to know if the AGP riser between the Prism and Onyx4 is the same. My understanding is that the Prisms used a slightly later generation of ATI cards than the Onyx4s did, but I'm not certain if there were other differences. Indy99, if you can check the part number on the one from that Prism, I can check the one on my Onyx4 so we can compare notes.