Onyx IR issues - Printable Version

+- IRIX Network Forums (//forums.irixnet.org)
+-- Forum: SGI/MIPS (//forums.irixnet.org/forum-3.html)
+--- Forum: Hardware/Triage/Repair (//forums.irixnet.org/forum-11.html)
+--- Thread: Onyx IR issues (/thread-2207.html)

---

Onyx IR issues - mosiniak - 06-10-2020

Hi,

I have some issues with my IR graphics. There is for sure at least one problem (there is no output signal). I ran irsaudit and got errors, but I'm not sure where the problem lies and what it is connected with.

First of all I ran the gfxinfo and irsaudit without any options and I got strange output from gfxinfo and error on irsaudit with DG. Then I ran irsaudit with -ge -rm and -dg options and each time I got error. During irsaudit -rm the was and error and Onyx was restarted. What do You think, are all these errors connected together and there is a single cause - e.g. DG board?

I attached all the results from irsaudit, gfxinfo and hinv.

---

RE: Onyx IR issues - jan-jaap - 06-10-2020

You have some truly insane numbers in your 'gfxinfo' output. 88 GE's,  205583996 RM's, 2623476 display channels ...

IR stores some config data in an eeprom on the GE board. I'd check out it's sanity first with ireeprom:  https://wiki.preterhuman.net/InfiniteReality:_The_Graphics_Engine_EEPROM_Commands

Also, IR requires the very last revision VCAM on the IO4. It must be part # 030-0500-206

---

RE: Onyx IR issues - mosiniak - 06-10-2020

(06-10-2020, 02:54 PM)jan-jaap Wrote:  You have some truly insane numbers in your 'gfxinfo' output. 88 GE's,  205583996 RM's, 2623476 display channels ...

No one has such a powerfull system  

(06-10-2020, 02:54 PM)jan-jaap Wrote:  IR stores some config data in an eeprom on the GE board. I'd check out it's sanity first with ireeprom:  https://wiki.preterhuman.net/InfiniteReality:_The_Graphics_Engine_EEPROM_Commands

I started ireeprom -i and got a sad result (see below). On the page you mentioned, I can read: "The graphics engine (GE) in Silicon Graphics SGI Onyx2 and SGI Onyx3 Infinite Reality (IR) stores the microcode in EEPROM". Do you know if this also applies to Onyx1?

(06-10-2020, 02:54 PM)jan-jaap Wrote:  Also, IR requires the very last revision VCAM on the IO4. It must be part # 030-0500-206

My IO4 is 030-0815-00x and VCAM is 030-0500-206 so this is not the reason.

---

RE: Onyx IR issues - ghost180sx - 06-14-2020

I expect you have tried removing the DG board and re-seating it? Are there any cables or connector boards that need to be checked?

Was this system working the last time you started it and this happened while it was sitting?

---

RE: Onyx IR issues - mosiniak - 06-15-2020

(06-14-2020, 05:21 AM)ghost180sx Wrote:  I expect you have tried removing the DG board and re-seating it? Are there any cables or connector boards that need to be checked?

Yes, I removed all the boards and very carefully inspected them, their connectors and connectors on the backplane (I was looking for bent pins). Everything was OK. I took photos and I will upload them soon. I checked also the triangle bus connector and it looks OK.

(06-14-2020, 05:21 AM)ghost180sx Wrote:  Was this system working the last time you started it and this happened while it was sitting?

No, I bought this system in December 2016. First, I kept it warm (at room temperature) for 6 months (to get rid of the stale smell), and then disassembled and cleaned the whole system (there were a lot of dead insects and spiders inside). Then after turning on the power the system worked (it was serial console output), but I could not get video output (there was a problem with the monitor). After some time, I bought a SOG-accepting monitor, but there was also no video output. There was also a hard disk failure and later I bought a new IBM HVD disk. In May 2020, I installed IRIX and ran irsaudit to see if something is wrong.


irsaudit can detect hardware:
but gfxinfo can't:
Any ideas, what I could be doing wrong?

---

RE: Onyx IR issues - jan-jaap - 06-16-2020

It's possible that you need to run ireeprom with graphics running (/usr/gfx/startgfx). Can't try right now -- I'm using the screen elsewhere...

irsaudit definitely needs to be run with exclusive access to the graphics hardware (/usr/gfx/stopgfx).

Both ireeprom and irsaudit work with all IR revisions, not only IR2 / IR3. They have been in IRIX since IRIX 6.2.

---

RE: Onyx IR issues - mosiniak - 06-16-2020

(06-16-2020, 07:29 AM)jan-jaap Wrote:  It's possible that you need to run ireeprom with graphics running (/usr/gfx/startgfx). Can't try right now -- I'm using the screen elsewhere...

I will check it. I didn't try running ireeprom with graphics running.

(06-16-2020, 07:29 AM)jan-jaap Wrote:  irsaudit definitely needs to be run with exclusive access to the graphics hardware (/usr/gfx/stopgfx).

I already noticed that.

(06-16-2020, 07:29 AM)jan-jaap Wrote:  Both ireeprom and irsaudit work with all IR revisions, not only IR2 / IR3. They have been in IRIX since IRIX 6.2.

Thanks for confirming that.

---

RE: Onyx IR issues - mosiniak - 06-18-2020

(06-16-2020, 07:29 AM)jan-jaap Wrote:  It's possible that you need to run ireeprom with graphics running...

You were right. With graphics running the ireeprom -i output is better, but there is problem still.


I have idea to move IO4+VCAM from rack to deskside machine with working RE2 graphics to see if IO4+VCAM are OK. Does it make any sense?

I also have another crazy idea - I have onyx deskside system with RE2 graphics and KONA ready backplane. I can move IR set to this machine (IO4+VCAM+GE+RM+DG), but there is a problem with power modules. IR graphics requires 3.3V supply (RE2 requires 5V). In deskside systems there is place for narrow (303) power modules, but in rack system I have wide (303x2) modules. What do You think is it possible to insert wide module into deskside system and left the cover open for a short time for tests?

What else could I check to find out what is going on?

---

RE: Onyx IR issues - CB_HK - 06-18-2020

(06-18-2020, 10:26 AM)mosiniak Wrote:  

(06-16-2020, 07:29 AM)jan-jaap Wrote:  It's possible that you need to run ireeprom with graphics running...

You were right. With graphics running the ireeprom -i output is better, but there is problem still.

Code:

/usr/gfx/stopgfx

/usr/gfx/KONA/bin/ireeprom -i
==== Pipe 0 ====
graphics not responding... trying again
graphics not responding... trying again
graphics not responding...loading ucode
ucode load complete...
Cannot communicate with graphics board

/usr/gfx/startgfx
Sent SIGHUP to xdm
IRIS 8# IR0: ARM: Welcome to ARMLand - 0/0x0d00
IR0: ARM: running...(sherwood-root 0310062200)
IR0: ARM: ******************************************************
IR0: ARM: * InfiniteReality/Reality Software, IRIX 6.5 release *
IR0: ARM: ******************************************************

/usr/gfx/KONA/bin/ireeprom -i
==== Pipe 0 ====
diag-map: Resource busy
WARNING: IR0: mopup timeout.
WARNING: IR0:  Fatal error.  bdata = 0xa800000000970000 death_reason = 17
Graphics FIFO depth is 26 words.

I have idea to move IO4+VCAM from rack to deskside machine with working RE2 graphics to see if IO4+VCAM are OK. Does it make any sense?

I also have another crazy idea - I have onyx deskside system with RE2 graphics and KONA ready backplane. I can move IR set to this machine (IO4+VCAM+GE+RM+DG), but there is a problem with power modules. IR graphics requires 3.3V supply (RE2 requires 5V). In deskside systems there is place for narrow (303) power modules, but in rack system I have wide (303x2) modules. What do You think is it possible to insert wide module into deskside system and left the cover open for a short time for tests?

What else could I check to find out what is going on?

Alright, take this with a grain of salt (and some caution). I have read that yes, you can in fact install the 303x2 modules in the desk side system. Yes, the back plate and cover have to be left off in order to do this. HOWEVER, I would wait until someone else either confirms this or says not to do it.

I'm looking for the post where I saw this written, but someone has tried this successfully before. If you're just doing it to test the boards I don't see it being an issue (specifically the rear of the machine being exposed).

---