SGI: Hardware

So I finally got around to putting in two vpro6's in my Octane.. they fit nicely and was far easier than I imagined. Each vpro6 had 2 screws to separate them in half.. and a 3rd smaller screw on the latch in order to disconnect it. I left the latch bits all there.. but of course they wont latch inside the octane (wouldnt be able to get them out then).

However, it did not work! The bottom vpro card is what it booted with. The top had no output. I ran hinv and gfxinfo while in that mode. Then, to make sure the vpro cards were not faulty or a connection not tight enough.. I removed the bottom vpro.. leaving the top one untouched, and rebooted. The top one worked fine then. So the connections are not the problem and the vpro's do not appear faulty.

With 2 vpro6's plugged in:





and with just 1 vpro6.. gfxinfo shows:



I did note that with 2 vpro's it seemed the Channel 0: did not show up in gfxinfo.. not sure what that was about.. or if its related?

Also, a hinv in the POST command line only shows 1 graphics card.
And the ODY32 boards have the same part numbers.

Anyone have any ideas? I am single CPU.. maybe SGI really meant it when they said it requires dual cpu for multi-head? Is a revision of one of my components too old? (Xbow is at 1.3?)

Thanks

1.3 Xbow revision is fine for V6 and V8, as stated on Ian's page. But it looks like JackLuminous correctly pointed out that in the install manual it is stated that 2 CPU's are required. So unless someone has a much more clever idea, your best bet is to get a cheap dual R10k@195MHz module and give it a go.

Last week, I put an additional V10 in my Octane2. A V12/DCD was already in there. It worked right away without any additional action. For once, I guess the dual CPU is a real requirement.

Jack Luminous wrote: Last week, I put an additional V10 in my Octane2

where did you found the dual-carrier ? and how much it was ?

ivelegacy wrote:

Jack Luminous wrote: Last week, I put an additional V10 in my Octane2

where did you found the dual-carrier ? and how much it was ?

I bought it from Nekochan user Mopar5150 for €200 (including shipping from across the pond). Very expensive and very hard to find. I waited for one (let alone at a good price) for years so I broke my SGI piggy bank this time and be done with it. It works great !

necron2600 wrote: Also, a hinv in the POST command line only shows 1 graphics card.
And the ODY32 boards have the same part numbers.

Anyone have any ideas? I am single CPU.. maybe SGI really meant it when they said it requires dual cpu for multi-head? Is a revision of one of my components too old? (Xbow is at 1.3?)

It's the fact that only one ODY board shows up which is a little strange. I suspect a dirty or damaged compression connector, or the PCB surface pads the compression connector mates to are dirty. If you change the VPro boards around, does it still see the same VPro board? If so, then you've got one VPro board with a dirty/broken compression connector. If not, then I'd suspect the frontplane. The frontplane PCB is often covered with a cake of fine black dust, built up over the years, and unless there was something in that slot before, it's also on the connector pads where your new card goes. I've seen this a couple of times when fitting a PCI cage to an Octane. To fix this I removed the frontplane and cleaned the connector pads with a cotton swab with some 97% alcohol.

To understand the requirement for dual CPUs, you have to understand the product placement back in the day: the dual head option was part of a "dual seat" option where two people would share a single Octane, but each have their own graphics head, keyboard and mouse (this requires a CADduo card). You can put up to 3 MGRAS graphics boards in an Octane, but I have never heard of the triple CPU option

FWIW: If needed, you can use a compression connector from any useless XIO option (FC, HVD SCSI, ...) to salvage another board, like a VPro graphics card.

in case, I have a V6 gfx for sale, 50 euro + shipping

necron2600 wrote: The bottom vpro card is what it booted with. The top had no output. I ran hinv and gfxinfo while in that mode. Then, to make sure the vpro cards were not faulty or a connection not tight enough.. I removed the bottom vpro.. leaving the top one untouched, and rebooted. The top one worked fine then. So the connections are not the problem and the vpro's do not appear faulty.

Interesting that either one works when the other is removed. My shoot from the hip reaction would be insufficient power delivery when both V6s are installed, and/or problems with the front plane connections.

As jan-jaap has already suggested environmental contamination between the V6s and the frontplane could contribute to either possibility.

If you're willing to retest/repost, the extra detail revealed in an hinv -vm (component revisions/hardware paths) might prove helpful, as could an hinv run at the prom monitor.

in my opinion: do not underestimate the compression connector and its catch mechanism
in my experience (with MENET ): the wrong catch mechanism did the difference between " properly working " and " not properly working "

Recondas may be right actually. Does your Octane have a cherokee power supply? (Silver handle) Or one of the older black handled ones? I might have lost my mind, but I thought I remembered that the output on the cherokee was significantly more...

-Jesse

In necron2600's hinv the part code for the power supply is listed:


So that is definitely a Cherokee. It's still possible that the Power supply has a fault and doesn't reach its max output.

I'll try again this weekend. I have a spare silver-handled PSU from another Octane to try too.

Regarding the locking mechanism.. I assume that is the latch on the Vpro cards. I do have those latches disconnected (single small screw at the pivoting point removed from each Vpro card which is required to take off the vpro from the rest of the board). And I am sliding in the vpro's by hand.. sometimes the angle is off and I have to pull it back out a bit and try again to connect straight. I know not to remove them when powered on.. as without the latches/locking it would come out easy with a tug.

When both vpro cards are in, the bottom one works.. and when I remove the bottom one, the top one works.. I would imagine that means the compression connectors/contacts are good?

A problem with power delivery does make sense... so we'll see how my spare PSU test goes.

Can anyone confirm that I should be able to see two graphics boards in the PROM monitor's hinv output? I'd rather test things there instead of booting up and shutting down all the time.

Thanks

dexter1 wrote: In necron2600's hinv the part code for the power supply is listed:

Code: Select all

PWR.SPPLY.ER Board: barcode AAE9050189 part 060-0035-001 rev  C

So that is definitely a Cherokee. It's still possible that the Power supply has a fault and doesn't reach its max output.

Ah yes, it would help if I was paying more attention! Given the age of some of these systems now (and the amount of time many of them have spent powered off before being revived by one of us), I wouldn't be surprised if the PS wasn't quite up to the task.

-Jesse

Update:
I replaced my power supply with another. Popped in two Vpro6s. hinv on bootup/PROM shows only 1 graphics card. Continued booting into IRIX and it also only shows 1 graphics card in hinv -vm and gfxinfo.

hinv with new powersupply and 2 vpros: (video only comes out of the bottom Vpro):

Location: /hw/node
PM10400MHZ Board: barcode LKH759 part 030-1475-001 rev C
Location: /hw/node/xtalk/15
IP30 Board: barcode GHZ615 part 030-0887-003 rev M
Location: /hw/node/xtalk/15/pci/2
FP1 Board: barcode 06909C part 030-0891-003 rev E
PWR.SPPLY.ER Board: barcode AAE9070128 part 060-0035-001 rev C
Location: /hw/node/xtalk/9
ODY32 Board: barcode LDE123 part 030-1403-003 rev B
1 400 MHZ IP30 Processor
Heart ASIC: Revision E
CPU: MIPS R12000 Processor Chip Revision: 3.5
FPU: MIPS R12010 Floating Point Chip Revision: 0.0
Main memory size: 896 Mbytes
Xbow ASIC: Revision 1.3
Instruction cache size: 32 Kbytes
Data cache size: 32 Kbytes
Secondary unified instruction/data cache size: 2 Mbytes
Integral SCSI controller 0: Version QL1040B (rev. 2), single ended
Disk drive: unit 1 on SCSI controller 0 (unit 1)
Integral SCSI controller 1: Version QL1040B (rev. 2), single ended
IOC3/IOC4 serial port: tty1
IOC3/IOC4 serial port: tty2
IOC3 parallel port: plp1
Graphics board: V6
Integral Fast Ethernet: ef0, version 1, pci 2
Iris Audio Processor: version RAD revision 12.0, number 1
PCI Adapter ID (vendor 0x10a9, device 0x0003) PCI slot 2
PCI Adapter ID (vendor 0x1077, device 0x1020) PCI slot 0
PCI Adapter ID (vendor 0x1077, device 0x1020) PCI slot 1
PCI Adapter ID (vendor 0x10a9, device 0x0005) PCI slot 3

Here is gfxinfo with new powersupply and 2 vpro's (but only bottom has output):
Graphics board 0 is "ODYSSEY" graphics.
Managed (":0.0") 1920x1200
BUZZ version A.1
PB&J version 1
32MB memory
Banks: 2, CAS latency: 3
Monitor 0 type: Unknown
Channel 0:
Origin = (0,0)
Video Output: 1920 pixels, 1200 lines, 60.02Hz (1920x1200_60)


And for FYI:
With old/original power supply, I noted the hardware location does change when only the top vpro is plugged into the Octane, and then when both vpros are plugged in (where the bottom vpro has output):

With top vpro only plugged in:
Location: /hw/node/xtalk/11
ODY32 Board: barcode LDZ638 part 030-1403-003 rev B

With both vpros plugged in (but video output only works on bottom vpro):
Location: /hw/node/xtalk/9
ODY32 Board: barcode LDE123 part 030-1403-003 rev B

I'll have to see about finding an Octane2 and go blue skins instead then.

Update (again

I got an Octane2. Had a v6 in it. Put in a 2nd v6 and .... it worked! Dual Head. Single CPU.

FYI, the hinv in PROM will show two ODYSSEY cards.

However, there is now another slight problem. Fastfan kicks in (probably because I have an XIO in the bottom slot). The system starts up relatively quiet and once it finishes its power on testing the fan(s) go fast. If I go back to a single V6 setup, the fans are quiet.
I have done plenty of searches on fastfan issues.. but setenv -p fastfan 0 or 1 , etc. dont seem to make any difference. (and it does show up in a printenv)

Note: This Octane2 seems a little bit quieter than my two green-skinned Octanes that have the exact same PSUs (when it has 1 vpro and fastfan is not on of course)

Anyone have any idea on fastfan and dual Vpro in an Octane? Maybe I missed something in my searches...

Thanks in advance,

Some more searching..
viewtopic.php?t=16730203
Looks like fastfan is going to be on and its normal with 2 VPros.. I'll just have to turn up some music?

Can you give us a hinv -vm please?

jan-jaap wrote: You can put up to 3 MGRAS graphics boards in an Octane, but I have never heard of the triple CPU option :mrgreen:

Octane's HEART chip was built to support up to four CPUs (and 16 DIMMs), but SGI never expanded the system out to actually use that. I've always wondered if the ramp-design of the HEART/BRIDGE heatsink was to allow for a quad module of sorts down the road.

necron2600 wrote:
Anyone have any idea on fastfan and dual Vpro in an Octane? Maybe I missed something in my searches...

Thanks in advance,

Strange because mine is not in fastfan mode despite a V12+V10. There was a fiber channel copper in place of the V10 before and no fastfan either.

Below is the hinv of this Octane2 system with dual Vpros and the fan noticeably faster/louder than if it had just 1 Vpro.

[octane]:~ $ hinv -vm
Location: /hw/node
PM10400MHZ Board: barcode LDV883 part 030-1475-001 rev B
Location: /hw/node/xtalk/15
IP30 Board: barcode KVY510 part 030-1467-001 rev D
Location: /hw/node/xtalk/15/pci/2
PWR.SPPLY.ER Board: barcode AAE1090141 part 060-0035-002 rev A
FP1 Board: barcode MBZ201 part 030-0891-003 rev H
Location: /hw/node/xtalk/11
ODY32 Board: barcode LLE123 part 030-1403-003 rev B
Location: /hw/node/xtalk/9
ODY32 Board: barcode LLX611 part 030-1403-003 rev B
1 400 MHZ IP30 Processor
Heart ASIC: Revision F
CPU: MIPS R12000 Processor Chip Revision: 3.5
FPU: MIPS R12010 Floating Point Chip Revision: 0.0
Main memory size: 2048 Mbytes
Xbow ASIC: Revision 1.4
Instruction cache size: 32 Kbytes
Data cache size: 32 Kbytes
Secondary unified instruction/data cache size: 2 Mbytes
Integral SCSI controller 0: Version QL1040B (rev. 2), single ended
Disk drive: unit 1 on SCSI controller 0 (unit 1)
Integral SCSI controller 1: Version QL1040B (rev. 2), single ended
IOC3/IOC4 serial port: tty1
IOC3/IOC4 serial port: tty2
IOC3 parallel port: plp1
Graphics board: V6
Graphics board: V6
Integral Fast Ethernet: ef0, version 1, pci 2
Iris Audio Processor: version RAD revision 12.0, number 1
PCI Adapter ID (vendor 0x10a9, device 0x0003) PCI slot 2
PCI Adapter ID (vendor 0x1077, device 0x1020) PCI slot 0
PCI Adapter ID (vendor 0x1077, device 0x1020) PCI slot 1
PCI Adapter ID (vendor 0x10a9, device 0x0005) PCI slot 3

Other than some newer revs, it looks largely the same as my old green-skin Octane, except that green-skin didnt work well with dual vpro for some reason.

Just to check to ensure that the noise I am hearing with dual vpro was indeed fastfan.. I had set fastfan mode to 1 in PROM, ensured it was set looking at it with printenv and powercycled, verified again its set with printenv. I then set it back to 0 and even unset it.

Thanks!