[weblacky]

Okay so I tried to button this up today and get it off my bench & done.  I tried the repaired PIMM (will post to that thread shortly) and I assembled EVERYTHING I've bought for the Fuel (full-up ALL PCI-X slots).  Results were...odd.

I placed in a Sound Blaster Audigy (SB0090) that I found for $5 at a recycler 3 weeks ago... Gives the expected boot registers error, but audio panel works so I assume it works (haven't placed speakers on it yet).

I put the flashed LSISAS3422X and a SATA 3.5" HDD in it...SAS ONLY no SATA, sucks... will be replacing it and maybe I will sell the HBA later on the forums.  I should have "read the manual" closer as it were.  I don't need SAS.  So bummer...

Put in an LSI 21320-IS U320 SCSI controller, does recognizing it but then sings the old "Song of the Irix Siren" --- FIRMWARE TOO OLD, update with lsfash.....UGHHGHGHGHHHH!!!  Does everything do this...come on!  So...now I have to pull THAT card and place it in a DOS machine and go through firmware update BS again...I'm glad I only need this in the Fuel and not my Tezros.

I placed an SGI-Mod-Flashed 1Gb NIC I had Jan-Jaap mod in a lot for me!  Was recognized instantly with driver so that was at least good as my deal with Jan-Jaap was like 1-2 years ago...then I had to throw them in a box and wait until I fixed my SGIs (ones that supported them anyway) to test.  So I took his word on the deal.  So far, that just worked...which was good!

So, more pulling, more firmware BS, sigh....So Fuel still works, ironing out the wrinkles before it goes on the shelf so I can move to the next SGI.


   

[jan-jaap]

Put in an LSI 21320-IS U320 SCSI controller, does recognizing it but then sings the old "Song of the Irix Siren" --- FIRMWARE TOO OLD, update with lsfash.....UGHHGHGHGHHHH!!!  Does everything do this...come on!  So...now I have to pull THAT card and place it in a DOS machine and go through firmware update BS again...

Nooo!

Read what it says. "lsflash" is in IRIX. In the later 6.5.x overlays at least.

[weblacky]

Put in an LSI 21320-IS U320 SCSI controller, does recognizing it but then sings the old "Song of the Irix Siren" --- FIRMWARE TOO OLD, update with lsfash.....UGHHGHGHGHHHH!!!  Does everything do this...come on!  So...now I have to pull THAT card and place it in a DOS machine and go through firmware update BS again...

Nooo!

Read what it says. "lsflash" is in IRIX. In the later 6.5.x overlays at least.

I found ZERO, ZIP, NOTHING, on lsflash (except under IBM, which wasn't even the same darn thing) online.  Unless you happen to have some sample/examples of how to use this...considering the firmware for this card is MPT  (I think?) and broken into several files...I'll stick with the company-directed way since this card isn't the SGI card anyway.

But While I have you, can you please pop over the PIMM repair thread?  I have some question that might be up your alley!

[weblacky]

Hi All,
I know I said I was done...but nagging issues kept me at bay here.  Right now I only need to go and get a right-angled SATA cable for the secondary HDD to test the new 1064 LSI controller in it...everything else has NOW BEEN Resolved.  There was one anomaly that happened once, I'll show that...otherwise...smooth sailing.

So big news. The original V10 seems to work!!!  I wrestled the replacement V10 out and placed the V10 that came with it that I replaced the DS1780 IC in and shorted that XIO Bias 12V short out on it (still never found the shorted I shorted out).
   


Now the anomaly, the first time I booted I was greeted with a USB_OHCI_INTR unhandled interrupt errors several times (booting and shutdown console):
   
   

But I then swapped in the NEW V10, no USB error, swapped back the repaired V10...no errors...shutdown and cold started twice with repaired V10 (after inserting new SATA controller.)...no USB errors...so I have no idea what happened there...but it's only happened once.

I had NOTHING in the USB ports so...don't know.  RAM is still holding fine at 2GB.

I am missing the DCD cover for my V10 so I'll cover it with tape in the meantime.  The old V10 frame is bent with the system case so it fits much better than the replacement anyway.  I was able to get ALL screws in with the old card (complete) where as I cannot align the video output screw holes with the new card.  So I'd prefer the old one...assuming it really works.

   

So that's a HUGE item, picture appears everytime I boot, recognizes the correct graphics specs...so for the time being..I'm declaring it working unless I see otherwise.  I'll keep the V10 I bought for troubleshooting since I have Tezros and it can be used in those for any reason.

I also placed the plastic face on and with the new latch, things work fine.

I do have a blue internal SCSI ZIP100 drive I was considering placing under the DVD drive, but I decided not to fiddle with that until I can better test the ZIP drive and I also still need a right-angled SATA cable to get into the VERY tight drive bay to attach an SATA disk to really see if the controller I just bought works or not (already upgraded it's firmware).

So now the the Fuel weighs like 45 lbs and has all it parts so I'm going to look into getting it off my work bench and proceed to the next round.


While I won't say all V10/12 cards can be repaired so easily, if any of you have Fuel/Tezro style XIO Graphics card that prevents booting.  You can easily get the DS1780 Datasheet and check for a short (using a multimeter in DIODE mode) between the sensor inputs (12v, 5v, 3.3v, etc...) and the GNDA and GNDD pins.  Also check between VCC and GNDA & GNDD.  I found basic shorts here.  The symptoms I saw was the L1 was missing nearly all acknowledgements for nodes and/or you're missing ALL the XIO ENV values.  If you get the first L1 symptom (missing acknowledgements for nearly ALL nodes) then just shutdown (remove power) and remove your Graphics card, then hook a serial terminal to your L1 port, get terminal working, then apply power to system (it will NOT be able to auto-power due to missing graphics) but check if the flood of "no acknowledgement from node" errors is now a trickle (nowhere near the amount you used to have).  Then do an ENV, if you have PSU 3.3v and 5V AND PIMM voltage...then your problem WAS/IS your DS1780 on your graphics card (at a minimum).  As you may know all the DS1780 ICs share a communication bus, a shorted IC will drag the entire bus down and none of them are able to talk, remove the shorted IC and they can talk again.

However that doesn't mean what they say is valid...you still may have DS1780 ICs that are misreporting (that still need change out) but at least they aren't disrupting the entire ENV bus.  So that's an easy way of isolating your ENV and graphics issue without buying parts/tools.

If you find this to be the case, I offer a DS1780 replacement service, and I can do a basic short check.  If you want an actual test on a system...that will cost extra (risk and all).  But the DS1780 is an easy way to do a non-invasive short test on these cards and give you more info.  I've not seen more samples so I don't know if this is unique or a trend.  Either way...it's free advice.  Check your DS1780 on your "dead" Fuel/Tezro graphics and look at the your ENV values with the card installed (offline L1, not auto-powered) and then removed.

For the V10, the DS1780 IC is here (highlighted in red).
   

I'll update the SAS thread when I have more.  

Also yes... I bought a RED AC cable from Amazon...because the Fuel is red and SGI should have extended the colors to their cables too :-)

[jan-jaap]

I will read over this entire thread when I have more time, but right now I just want to say well done!

[weblacky]

I will read over this entire thread when I have more time, but right now I just want to say well done!

Thanks! That means a lot from a Pro, like yourself. We build on the work of others and people such as yourself did a heck of a lot of work before I came along.  I'd not be able to even do any of this without your findings over the years, that you shared online, along with others.  Each putting together the bits to draw new theories or conclusions.

I do appreciate that SGI tech was made during the transition to very high speed frequencies and so most of the parts aren't BGA and can be worked by hand!  While there are some BGA parts, they don't make up the majority.  With the nice tools we have today it makes this doable (to a point).

I have also found SGI placed a heck of a lot of info on their test points and their firmware for board testing/troubleshooting (which I find odd because to my knowledge they didn't refurb boards..so they must have done it purely for Q&A and failure analysis).  Which helped somewhat.

[megaimg]

Hi All,
I know I said I was done...but nagging issues kept me at bay here.  Right now I only need to go and get a right-angled SATA cable for the secondary HDD to test the new 1064 LSI controller in it...everything else has NOW BEEN Resolved.  There was one anomaly that happened once, I'll show that...otherwise...smooth sailing.

So big news. The original V10 seems to work!!!  I wrestled the replacement V10 out and placed the V10 that came with it that I replaced the DS1780 IC in and shorted that XIO Bias 12V short out on it (still never found the shorted I shorted out).



Now the anomaly, the first time I booted I was greeted with a USB_OHCI_INTR unhandled interrupt errors several times (booting and shutdown console):



But I then swapped in the NEW V10, no USB error, swapped back the repaired V10...no errors...shutdown and cold started twice with repaired V10 (after inserting new SATA controller.)...no USB errors...so I have no idea what happened there...but it's only happened once.

I had NOTHING in the USB ports so...don't know.  RAM is still holding fine at 2GB.

I am missing the DCD cover for my V10 so I'll cover it with tape in the meantime.  The old V10 frame is bent with the system case so it fits much better than the replacement anyway.  I was able to get ALL screws in with the old card (complete) where as I cannot align the video output screw holes with the new card.  So I'd prefer the old one...assuming it really works.



So that's a HUGE item, picture appears everytime I boot, recognizes the correct graphics specs...so for the time being..I'm declaring it working unless I see otherwise.  I'll keep the V10 I bought for troubleshooting since I have Tezros and it can be used in those for any reason.

I also placed the plastic face on and with the new latch, things work fine.

I do have a blue internal SCSI ZIP100 drive I was considering placing under the DVD drive, but I decided not to fiddle with that until I can better test the ZIP drive and I also still need a right-angled SATA cable to get into the VERY tight drive bay to attach an SATA disk to really see if the controller I just bought works or not (already upgraded it's firmware).

So now the the Fuel weighs like 45 lbs and has all it parts so I'm going to look into getting it off my work bench and proceed to the next round.


While I won't say all V10/12 cards can be repaired so easily, if any of you have Fuel/Tezro style XIO Graphics card that prevents booting.  You can easily get the DS1780 Datasheet and check for a short (using a multimeter in DIODE mode) between the sensor inputs (12v, 5v, 3.3v, etc...) and the GNDA and GNDD pins.  Also check between VCC and GNDA & GNDD.  I found basic shorts here.  The symptoms I saw was the L1 was missing nearly all acknowledgements for nodes and/or you're missing ALL the XIO ENV values.  If you get the first L1 symptom (missing acknowledgements for nearly ALL nodes) then just shutdown (remove power) and remove your Graphics card, then hook a serial terminal to your L1 port, get terminal working, then apply power to system (it will NOT be able to auto-power due to missing graphics) but check if the flood of "no acknowledgement from node" errors is now a trickle (nowhere near the amount you used to have).  Then do an ENV, if you have PSU 3.3v and 5V AND PIMM voltage...then your problem WAS/IS your DS1780 on your graphics card (at a minimum).  As you may know all the DS1780 ICs share a communication bus, a shorted IC will drag the entire bus down and none of them are able to talk, remove the shorted IC and they can talk again.

However that doesn't mean what they say is valid...you still may have DS1780 ICs that are misreporting (that still need change out) but at least they aren't disrupting the entire ENV bus.  So that's an easy way of isolating your ENV and graphics issue without buying parts/tools.

If you find this to be the case, I offer a DS1780 replacement service, and I can do a basic short check.  If you want an actual test on a system...that will cost extra (risk and all).  But the DS1780 is an easy way to do a non-invasive short test on these cards and give you more info.  I've not seen more samples so I don't know if this is unique or a trend.  Either way...it's free advice.  Check your DS1780 on your "dead" Fuel/Tezro graphics and look at the your ENV values with the card installed (offline L1, not auto-powered) and then removed.

For the V10, the DS1780 IC is here (highlighted in red).


I'll update the SAS thread when I have more.  

Also yes... I bought a RED AC cable from Amazon...because the Fuel is red and SGI should have extended the colors to their cables too :-)

Great Job!