Nekochan Net

All, I've been investigating this for 3 months now, I got this Octane2 with V10 and stuff,
and it would die randomly, then I discovered it was mostly happening during compilation (heavy use of memory, disk ?) - compiling fltk-1.3.0 (cheers hamei - I've been secretly compiling dillo) is by far the easiest way to crash the machine.



I've tested this XBOW with FOUR different motherboards with the same results - all would die with the same error.
The same FOUR MB's do not crash in a fully working chassis with the same XBOW version.

I suspected it might have been related to PCI cage I got, or RAM, or even suspected SCSI controller - but 4 MB's (each had own ram) kind of disprove MB's fault...

Technically it's weird the machine would be "partially" broken - but who knows what other functions XBOW has besides switching XIO...

I don't think there's much I can do anymore nor do I have time to play with it.

It's free if anybody wants it for surgery/experiments/whatever (appropriate post is under hw for sale/trade with more details on pickup/shipping etc)...

Cheers

_________________
[click for links to hinv] JP: [ :Fuel: :Octane2: :O2: :Indy: ] PL: [ :Fuel: ]
For Sale: 2*O200 M/B, 2*O200 PSU, 6*256MB O200 RAM, 2*O200 SCSI Backplane, 2*O200 MSC, DMediaPro DM-2 ( 030-1653-002 Rev. H , XT-DIGVID) with Octane XIO pull (Origin pull optionally available)

But that's exactly what it's complaining about: an XIO message (XTalk packet) being bogus. The problem could reside in the buffer memory on the XBow itself, and there is not much you can do (unless you are a soldering gun guru) to fix this.

Actually, since the problem seems to only occur when there is a lot of activity occuring, tinkering with the widget configuration to reduce speed and/or burst sizes might help, but I have no idea whether this can be tuned from IRIX.

_________________
R4000 R4400 R4400 R8000 R10000 R4000PC R4000SC R5000SC R5000 RM7000 2xR10000 R12000 - 2x2xR10000 R16000
among more than 150 machines : Apollo, Be, Data General, Digital, HP, IBM, MIPS before SGI, Motorola, NeXT, SGI, Solbourne, Sun...

Now that I think about it, I recall that network performance was somewhat slower than on another octane2 - I might undig it from the grave and see whether there are other indicators (and look at xbstat more) - it won't bring it back, but might help others with diagnosis of similar issues.
Buffer memory does sound likely, and it might cause the issue only when the buffers go over some certain watermark.

Cheers

_________________
[click for links to hinv] JP: [ :Fuel: :Octane2: :O2: :Indy: ] PL: [ :Fuel: ]
For Sale: 2*O200 M/B, 2*O200 PSU, 6*256MB O200 RAM, 2*O200 SCSI Backplane, 2*O200 MSC, DMediaPro DM-2 ( 030-1653-002 Rev. H , XT-DIGVID) with Octane XIO pull (Origin pull optionally available)

So I came up with another idea which didn't help either - an XBOW fan, I thought that since it freezes only during "heavier tasks" it might be the chip overheating, I took the broken frontplane apart and tested that fan - (un)fortunately it spins, so overheating ruled out

_________________
[click for links to hinv] JP: [ :Fuel: :Octane2: :O2: :Indy: ] PL: [ :Fuel: ]
For Sale: 2*O200 M/B, 2*O200 PSU, 6*256MB O200 RAM, 2*O200 SCSI Backplane, 2*O200 MSC, DMediaPro DM-2 ( 030-1653-002 Rev. H , XT-DIGVID) with Octane XIO pull (Origin pull optionally available)

You could try pumping a thin spray of liquid nitrogen at the board ... don't forget to open the window

Did you ever have any further luck with the Dillo Experiment ?

Dillo's still on my todo list but I just installed a Dual Channel Display and a DMediaPro board and now the suckin' thing won't boot...

Hmm, I think I'm cursed or something, the XBOW 1.3 I used from another Octane is also dead, I haven't used it for some time, and now that I assembled it all back I only get the "power" light on the front status leds which is quite weird...

_________________
[click for links to hinv] JP: [ :Fuel: :Octane2: :O2: :Indy: ] PL: [ :Fuel: ]
For Sale: 2*O200 M/B, 2*O200 PSU, 6*256MB O200 RAM, 2*O200 SCSI Backplane, 2*O200 MSC, DMediaPro DM-2 ( 030-1653-002 Rev. H , XT-DIGVID) with Octane XIO pull (Origin pull optionally available)

Oh damn, vishnu. Didn't they tell you ? the DCD won't work with that setup.

But since you are a really nice guy and I hate to see you struggle like that, tell ya what. I'll take that dcd thing off your hands and your computer will boot again. Whatcha say, fifty or a hunnerd bucks sounds about right ? And shipping, of course. I wouldn't want to screw you on the shipping ....

Bwahh ha ha nice try hamei but my understanding is that if you want the DM2 in your 'tane to talk to the DM5 in your VBOB you have to use the much-dreaded DCD or your video out will be sorely alack! Now, where am I wrong on this unfounded contention?

Nothing ventured, nothing gained ...

I'm trying, so far with very little success, to set something like this up:

^ Looking at that diagram, I see something peculiar. They have the graphics monitor connected to the main 13w3 V12 output. But that has never worked for me with a DCD. When I tried it years ago, that output went flicker flicker flicker flip flop jump jump but no usable display. Are you using one of the 2@ graphics settings ?

You could ask some other dcd owners to verify this, mine is currently sitting in a drawer.

Alternatively, you could try connecting your graphics output to the other dcd output port and see what happens ....

That's my understanding too, in fact in the documentation SGI says to screw the cover plate over the 13w3 connection once the DCD has been installed, and I seem to recall someone else pointing out that that diagram is wrong for that same reason in another thread where it was posted. But anyway, I'm stuck way short of either of those points, my machine hasn't been able to boot since I put the DCD and DM2 boards in. Just have to get up the gumption to pull the V12 and try installing them one at a time instead of both at once, I guess...

Connecting a serial terminal so you can look at the power-on diagnostics might save you some of that repetitive screwdriver exercise.

_________________
***********************************************************************
Welcome to ARMLand - 0/0x0d00
running...(sherwood-root 0607201829)
* InfiniteReality/Reality Software, IRIX 6.5 Release *
***********************************************************************

You guys and your serial consoles ... why do you always want to do everything the easy way ?