[jwhat]

Hi SGi'ers,

I decided to boot up my SGI boxes the other day to see if they where all working ok and I have got an issue with my Octane2 V12.

The machines boots, but every time it goes into graphics (startgfx) the screen goes blank and I get no graphics, so its an SNGI Octane2.

I have done some checking including:

1. Tested LCD on another computer at same resolution (1600x1200 @ 60) - ok
2. Cleared /var/X11/Xvc/Odyssey0_TimingTable to force it to go to default resolution - still get blank graphics
3. Booted into "single" mode and did a reset of monitor resolution /usr/gfx/setmon -x 1600x1200_60 - reports ok but on reboot still blank
4. Pulled out entire XTALK assembly and reseated it and verified that all card are identifying ok via hinv -mv - all ok but still blank graphics on reboot
5. Verified that X Window (xdm & graphics) have started ok, by connecting to it from another machine "X -query MY-BOX" - all ok I can use machine remotely
6. Tested with an alternate LCD (SGI F180 I think) - same problem

So it seems that something has gone wrong with VPro V12.
This one also has DCD in it and was all running ok (6 or 12 months ok... can't remember when I last used it)

So I am looking for the mythical ODYSSEY/VPRO Diagnotics Disk so I can run diagnostic on my machine (there is no /usr/diags directory)

This is likely only community that will have this (Ian M any chance you have one of these ...)

As last step I will remove the DCD and try with analog cable to see if this helps.

On another note I have put Linux L3 / L2 Simulator CD here: https://www.graphica.com.au/files/CD-IST_3.24.tar.gz , I think this should be added into IRIX downloads as it is helpful for people with O300/300 series machines.

Thanks in advance for any assistance.

Cheers from OZ:

jwhat

[jan-jaap]

I have the IST 2.7 CD which has Octane/ODYSSEY diagnostics. But it's from 2002 so there's a chance it only works with ~ IRIX 6.5.15

Would love to have a newer version ...

[Jacques]

I have the IST 2.7 CD which has Octane/ODYSSEY diagnostics. But it's from 2002 so there's a chance it only works with ~ IRIX 6.5.15

Would love to have a newer version ...

I was looking for this a while ago as well to test my v8, sorted my problem though as it was down to the multiple OpenGL instance bug on v6/v8.

jwhat - I'd pull the dcd first as a test.

[jwhat]

Hi Jan-Jaap & JacquesT,

i pulled DCD and tested with RGB Cable (finding this in garage was the hardest part of testing ...) and graphics came up ok.

I noticed that DCD board looked like it had signs of fluid leakage near "Intel Flash" (details covered by sticker with labelling: "1591 / 2 4302 / C9088") .

So cleaned board with alcohol and swaps and re-added it and V12 with DCD now working ok.

Does anyone know details of Intel chip and is this known to fail ?

BTW Jaques I also get console messages of having multiple sgifd servers running, but this does not appear to be preventing anything from working and doing ps -A does not show any sgifd processes at all.

Jan-Jaap for future references could I get download of diagnostic image so I see if this works with V12 & 6.5.29 (I have never bothered to update this machine to 6.5.30)

BTW here is TimingTable for SAFE mode:

> # cat Odyssey0_TimingTable.safe
> 1280x1024_60_safe
> 16
> 1

Thanks for tips and suggestions.

Cheers from Oz,

jwhat.

[Jacques]

Hi Jan-Jaap & JacquesT,

i pulled DCD and tested with RGB Cable (finding this in garage was the hardest part of testing ...) and graphics came up ok.

I noticed that DCD board looked like it had signs of fluid leakage near "Intel Flash" (details covered by sticker with labelling: "1591 / 2 4302 / C9088") .

So cleaned board with alcohol and swaps and re-added it and V12 with DCD now working ok.

Does anyone know details of Intel chip and is this known to fail ?

BTW I Jaques I also got console messages of having multiple sgifd servers running, but this does not appear to be preventing and doing ps -A does not show any sgifd processes at all.

Jan-Jaap for future references could I get download of diagnostic image so I see if this works with V12 & 6.5.29 (I have never bothered to update this machine to 6.5.29)

BTW here is TimingTable for SAFE mode:

> # cat Odyssey0_TimingTable.safe
> 1280x1024_60_safe
> 16
> 1

Thanks for tips and suggestions.

Cheers from Oz,

jwhat.

Hey, glad you got it working!

[jwhat]

Hi Jan-Jaap & JacquesT,

seems I got lucky last time as Octane V12 appears to have died again ...

I removed the DCD (again) and did clean of board and replaced, still not working.

So I removed DCD and was able to get graphics console via "single" mode.

So I revisited the IST V12 diagnostics disk.

This time I did clean IRIX 6.5.22 install and then installed the IST odyssey 6.5.15 distribution from the IST 2.7 CD ...

I had to write a little shell script to: cd to diags directory, run the diagnostics and then shutdown the machines (as I lose UI after running diagnostics and so was unable to confirm if they had completed or not):

The diagnostics appeared to work with clean IRIX install and completed to point of coming up with error:

>> % cat details.log
>> TEST registers      The Register test                      Test(1/3), Loop(1/1)
>> INFO                Saving current register values
>> INFO                Walking 1's
>> INFO                Walking 0's
>> INFO                Restoring original register values
>> RSLT registers      PASS        Board#0: Odyssey Register Test PASSED       
>> TEST cfifomem      The sdram CFIFO Memory test            Test(2/3), Loop(1/1)
>> INFO                Cfifo bypass already on...
>> RSLT cfifomem      PASS        Board#0: Sdram CFIFO Memory Test PASSED     
>> TEST dcb            The DCB Bus Test                      Test(3/3), Loop(1/1)
>> INFO                Saving current register values
>> INFO                Walking 1's
>> INFO                Walking 0's
>> INFO                Restoring original register values
>> RSLT dcb            PASS        Board#0: Odyssey DCB Test PASSED             
>> TEST i2c            The I2C Data Bus Test                  Test(1/1), Loop(1/1)
>> INFO                Saving current register values
>> INFO                Walking 1's
>> **** ERROR 003006  PBJ_I2C_opt_control exp 0x1 recv 0xff
>> 
>> INFO                Restoring original register values
>> RSLT i2c            FAIL        Board#0: errcode==I2CBUS                     
>> INFO                Maximum error count (1) reached

Not sure if any one else has run diagnostics on failed V12.

I believe that there have been a number of V12 failures (especially in Tezro/Fuels).

For now it looks like I will have to replace board and then maybe I can have a closer look at the failing one to see if there is any obvious issues (leaky capacitor or burnt out component or such).

Another task requiring Weblacky diagnotics skills !

Cheers from Oz,


jwhat/John.

[weblacky]

This topic is on my “after PSUs” list. It would be nice to find the i2c bus and find if something is clamping it or dead. I’m sure Jan-Jaap has likely looked into graphics failures to have some clue where these components might be.

Rumor has it that on at least fuel and Tezro the ds1780 monitoring chip was used via i2c to monitor something on high-end graphics. Perhaps there’s a relationship? Unknown of Octane v12 is anything like newer station V12, I have both those things but never bothered to look at them!

It will eventually take some time but an i2c failure may only indicate the inability to talk to a single i2c bus with multiple devices on a single chain.

I had fully planned to ask people for “dead” SGI graphics cards and perform a side-by-side comparison on my Huntron to see if there is an actual short/impedance change on the bus lines and try to track it down.

So I guess to anyone out there that has i2c related error messages, please stay with us on the forums for when the call goes out for research samples of these failures!

After PSUs, high-end graphics seem to be the next pain point, that being said (outside i2c) I doubt we’ll be able to do much for like overheating damage on these old graphics card (not that I think that’s related in any way to i2c). In the back of my mind I cannot help but wonder if we need a member with CNC & milling expertise to create rigid frames that replace or attach in a complimentary way to hotter-running graphics cards to stave off failure due to bending on thermal cycling or cooking nearby passives.

Basically the same issue with modern graphics cards still existed back then (sad). When pushing the thermals, physical failures increase. It would be nice to bandaid what we have with extra fans or heat sinks.

[jwhat]

Hi Weblacky,

good to hear that graphics are high in your priorities.

After all it was the Graphics in SGI that formed heart of the company (thanks Jim Clark ;-) ).

I have sourced a replacement V12 , so once that arrives I will take closer look at dead one and see if I can get heat sink etc off.

In email exchange with mopar5150, he said had recovered a DCD by having it “re-flowed”:

“The DCDs die because of heat. I have recovered one in the past by having it "re-flowed" in an oven at a board house.”

It might be that V12 could be revived by same treatment.

I will likely revisit in a month once replacement arrives.

Thanks again.

Cheers from Oz,


jwhat/John

[Geoman]

Can someone mirror the file CD-IST_3.24.tar.gz? It downloads with less than acoustic-coppler speeds.

Concerning the death of DCDs, I guess I have to glue heatsinks on the chips then. What chips are going to be hot? The FPGA and the 2 "Analog Devices" ASICS I guess. What passive components could die? the Tantalums? Then on the other hand - at least Tezro has much airflow over these chips

[jwhat]

Can someone mirror the file CD-IST_3.24.tar.gz? It downloads with less than acoustic-coppler speeds.

Concerning the death of DCDs, I guess I have to glue heatsinks on the chips then. What chips are going to be hot? The FPGA and the 2 "Analog Devices" ASICS I guess. What passive components could die? the Tantalums? Then on the other hand - at least Tezro has much airflow over these chips

Hi Geoman,

if you are referring to the link I posted in the original post, sorry I am on the end of HFC connection in Oz so downloads is quite a long way away.

I believe that the L2/L3 download is now also available from Irix Network files area.

This download does not include the V12 (on IST 2.7 CD) diagnostics which I got from Jan-Jaap.

It would be good if this could also be make the IST 2.7 CD more generally available... maybe via CD-ROM Archive: https://archive.org/details/cd-roms

Cheers from Oz,

jwhat/John