[LarBob]

Now I'm puzzled. Do these PPC970 systems use 68pin SCSI disks in their drive sleds instead of 80pin SCA SCSI disks the POWER5(+) systems use?

Correct, they are just regular 68-pin SCSI. You can even see this in the photo here: https://c1.staticflickr.com/8/7219/73809...9f86_b.jpg

Sorry, where do I have to look in that photo to see that?

http://www.ibmfiles.com/pages/intellipower185.htm
According to ibmfiles the GXT 4500p should be compatible.

The GXT4500P and GXT6500P are both supported and compatible. 7037 and 7047 use the exact identical motherboard, same FRU. Feel free to even look up the FRUs on IBM's website and you'll see they're shared for 7037/7047:
42R5210*
42R5208**
https://www.ibm.com/support/knowledgecen...system.htm

I don't doubt that the hardware of 7037 and 7047 is identical. But does same FRU also mean same firmware? I tought about some artificial firmware (or maybe VPD - if these machines also have something like that) limitation in the 7037.

As Larbob wrote that his GXT4500 is known working, but doesn't work in a 7037 - I interpolated: (1) not working in any compatible slot, (2) neither in SMS nor later and (3) on the same display where it worked earlier with another machine - whereas it should work in a 7047 according to the IBM documentation, the "conclusion" that there could be a limitation in the 7037's firmware, looked straight-forward to me. But maybe that's too far fetched...

Okay, I got AIX 7.1 installed on a 68pin disk that arrived today. So yes, AIX 7.1 does work!

Code:

AIX Version 7
Copyright IBM Corporation, 1982, 2010.
Console login: root
*******************************************************************************
*                                                                             *
*                                                                             *
*  Welcome to AIX Version 7.1!                                                *
*                                                                             *
*                                                                             *
*  Please see the README file in /usr/lpp/bos for information pertinent to    *
*  this release of the AIX Operating System.                                  *
*                                                                             *
*                                                                             *
*******************************************************************************

# lscfg -vp | grep proc
 proc0            U789B.001.1005EAA-P1            Processor

Noteworthy: this error during startup.

Code:

Automatic Error Log Analysis for sysplanar0 has detected a problem.
The Service Request Number is
  DA180013: Refer to the Error Code to FRU Index in the system service guide..

Haven't found anything on this # online and didn't find a copy of the service guide in a cursory search.

[micrex22]

Now I'm puzzled. Do these PPC970 systems use 68pin SCSI disks in their drive sleds instead of 80pin SCA SCSI disks the POWER5(+) systems use?

Correct, they are just regular 68-pin SCSI. You can even see this in the photo here: https://c1.staticflickr.com/8/7219/73809...9f86_b.jpg

Sorry, where do I have to look in that photo to see that?

I don't doubt that the hardware of 7037 and 7047 is identical. But does same FRU also mean same firmware? I tought about some artificial firmware (or maybe VPD - if these machines also have something like that) limitation in the 7037.

As Larbob wrote that his GXT4500 is known working, but doesn't work in a 7037 - I interpolated: (1) not working in any compatible slot, (2) neither in SMS nor later and (3) on the same display where it worked earlier with another machine - whereas it should work in a 7047 according to the IBM documentation, the "conclusion" that there could be a limitation in the 7037's firmware, looked straight-forward to me. But maybe that's too far fetched...

You can see the characteristic 68-pin rainbow cables (fairly famous adaptec cables), and it's possible to see some of the 68-pin cable heads (albeit it's a dark photo). It's pretty easy to spot if you deal with 68-pin SCSI often I suppose. For instance at a glance you can tell this is 68-pin SCSI as well:


If the firmware was different.. there would have to be a different FRU. Otherwise you'd be getting the wrong firmware boards for both systems with no way to tell them apart... not to mention IBM ITSO would have no way of replacing the proper botherboard with the appropriate firmware then.
IBM FRUs always retain the same characteristics. If there's something different (even as something as trivial if one uses leaded solder versus unleaded) there has to be a different FRU.

The GXT4500 will work on 42R5210/42R5208; there is no debate. If there's a problem it's unrelated to firmware as all of the motherboards have the same firmware. IBM took 7037 and threw it in a different deskside chassis to sell as an "IntelliStation", but all of the IntelliStation POWERs are actually motherboards borrowed from something else and thrown in a different chassis. They're not going to waste R&D on making something completely new each time.

Not to mention, the 7047 can utilize PowerVM just like 7037 because of their identical firmware. If that's not enough to convince you, 7047 even has the same firmware to download as 7037:
https://www.ibm.com/support/knowledgecen...rm_atx.htm

[johnnym]

Now I'm puzzled. Do these PPC970 systems use 68pin SCSI disks in their drive sleds instead of 80pin SCA SCSI disks the POWER5(+) systems use?

Correct, they are just regular 68-pin SCSI. You can even see this in the photo here: https://c1.staticflickr.com/8/7219/73809...9f86_b.jpg

Sorry, where do I have to look in that photo to see that?

I don't doubt that the hardware of 7037 and 7047 is identical. But does same FRU also mean same firmware? I tought about some artificial firmware (or maybe VPD - if these machines also have something like that) limitation in the 7037.

As Larbob wrote that his GXT4500 is known working, but doesn't work in a 7037 - I interpolated: (1) not working in any compatible slot, (2) neither in SMS nor later and (3) on the same display where it worked earlier with another machine - whereas it should work in a 7047 according to the IBM documentation, the "conclusion" that there could be a limitation in the 7037's firmware, looked straight-forward to me. But maybe that's too far fetched...

You can see the characteristic 68-pin rainbow cables (fairly famous adaptec cables), and it's possible to see some of the 68-pin cable heads (albeit it's a dark photo).

Thanks, I now already found this one - titled "185_power_hdd_bay" - on your website, which makes things pretty clear and if you would have pointed me to that, it also would have done so earlier already:



From other pictures from the web showing a 7037-A50 - e.g. like this one here:



...it looked like that black cover could cover a hot-plug drive cage. It looks pretty similar to the covers used on e.g. p5 520, although it's a little wider now that we can compare them "side by side":



Hence I initially assumed the 7037-A50 would use the same disk sleds as the other p5 systems and a SCSI backplane - in the end it was sold as System p5 185.

It's pretty easy to spot if you deal with 68-pin SCSI often I suppose. For instance at a glance you can tell this is 68-pin SCSI as well:

FWIW internal 68pin SCSI cables alone are not significant enough in my opinion, because the cables from SCSI controllers to SCSI backplanes are often just 68pin SCSI cables, though maybe usually without cable heads between the ends.

If the firmware was different.. there would have to be a different FRU. Otherwise you'd be getting the wrong firmware boards for both systems with no way to tell them apart...

See, that was my question, thanks for the clarification.

The GXT4500 will work on 42R5210/42R5208; there is no debate. If there's a problem it's unrelated to firmware as all of the motherboards have the same firmware. IBM took 7037 and threw it in a different deskside chassis to sell as an "IntelliStation", but all of the IntelliStation POWERs are actually motherboards borrowed from something else and thrown in a different chassis. They're not going to waste R&D on making something completely new each time.

I specifically referred to the firmware (not the hardware), which could have been different for different applications - at least I've seen such with other vendors.

Not to mention, the 7047 can utilize PowerVM just like 7037 because of their identical firmware. If that's not enough to convince you, 7047 even has the same firmware to download as 7037:
https://www.ibm.com/support/knowledgecen...rm_atx.htm

Please, I never questioned your statements, or did I? I think I explained already why I came to my conclusion regarding the nonfunctioning of a known good graphics card in a specific machine type earlier.

EDIT, to make it sound less harsh maybe :
Dear micrex22, you seem to have a profound knowledge of IBM machinery, I don't question that. Please forgive me that I have to resort to my intuition and experience with other machines and vendors when trying to diagnose issues with machines I don't have available for verification. This can sometimes direct one to conclusions that might be proved wrong later. In this case a firmware limitation sounded conclusive, given the fact that IBM is known to limit hardware capabilities like number of usable CPU cores, etc.. Maybe next time you could just jump in a little earlier with some facts?  

[LarBob]

The GXT4500 will work on 42R5210/42R5208; there is no debate. If there's a problem it's unrelated to firmware as all of the motherboards have the same firmware. IBM took 7037 and threw it in a different deskside chassis to sell as an "IntelliStation", but all of the IntelliStation POWERs are actually motherboards borrowed from something else and thrown in a different chassis. They're not going to waste R&D on making something completely new each time.

Right, so it should work. I wonder what the problem is here. Once again, I've verified that it still works in a 43p machine I have. I found a GXT135p for cheap which I'm gonna try as soon as it arrives.

Edit: OK, the code from the front panel when the system hangs during 5.3 installer boot is 0c45, meaning `Cannot configure the console.` I've selected the console in SMS and such. I think I'm going to eliminate the Serial->USB adapter in between and make sure there is *no flow control*.

Edit 2: No success. I actually think the problem is that this 5.3 is too old. I have LCD4-7463-04, but you need LCD4-7463-05 or LCD4-7444-01 AKA 53E at least for this machine it seems.

Edit 3: Got 5.3E.

Code:

AIX Version 5.3


******* Please define the System Console. *******

Type a 1 and press Enter to use this terminal as the
  system console.
Pour definir ce terminal comme console systeme, appuyez
  sur 1 puis sur Entree.
Taste 1 und anschliessend die Eingabetaste druecken, um
  diese Datenstation als Systemkonsole zu verwenden.
Premere il tasto 1 ed Invio per usare questo terminal
  come console.
Escriba 1 y pulse Intro para utilizar esta terminal como
  consola del sistema.
Escriviu 1 1 i premeu Intro per utilitzar aquest
  terminal com a consola del sistema.
Digite um 1 e pressione Enter para utilizar este terminal
  como console do sistema.

Yay!

EDIT: Got graphical output as well with a GXT 135p.

[johnnym]

EDIT: Got graphical output as well with a GXT 135p.

Oha, so a GXT135P works in your p5 185.

Any chance you get your hands on another known-good GXT4500P or GXT6500P for verification?

And just in case the used slot is special, does your GXT4500P maybe work in the slot you're currently using your GXT135P in?

And depending on how you used the DVI output on your GXT4500P, have you ever tried to use it with a DVI to VGA adaptor?

[LarBob]

EDIT: Got graphical output as well with a GXT 135p.

Oha, so a GXT135P works in your p5 185.

Any chance you get your hands on another known-good GXT4500P or GXT6500P for verification?

And just in case the used slot is special, does your GXT4500P maybe work in the slot you're currently using your GXT135P in?

And depending on how you used the DVI output on your GXT4500P, have you ever tried to use it with a DVI to VGA adaptor?

I could get another GXT4500p but I don't really want to as this one already works in my 43p, just not in the 7037-a50.

No, I can't get the GXT 4500p to work at all in the machine. I have it in the proper slot which is the same one I have the GXT 135p in now.

I just used straight DVI-DVI. In any case, that almost certainly isn't the issue as the 4500p doesn't even get listed in the diagnostics.

[johnnym]

Just stumbled upon this by chance and thought it could fit in here:

The CoD (capacity-on-demand) information available in the ASMI of my System p5 520Q has the following intriguing bit not set:

Code:

CoD Capability Settings
Capability bit information             Capability bit status/count
[...]
6 3D work-station graphics card OK     False
[...]

I have to check what my POWER 285 tells me for this bit, but the description seems to speak for itself. If this also applies to the p5 185 is the question, though.

[micrex22]

I could get another GXT4500p but I don't really want to as this one already works in my 43p, just not in the 7037-a50.

No, I can't get the GXT 4500p to work at all in the machine. I have it in the proper slot which is the same one I have the GXT 135p in now.

I just used straight DVI-DVI. In any case, that almost certainly isn't the issue as the 4500p doesn't even get listed in the diagnostics.

I have an update in regards to this, but my misfortune may explain what's going on. I blindly went ahead and bought two GXT4500Ps for my 185 (because I wanted dual monitor support). However... it appears there are different models of GXT4500P and I was not paying attention to FRUs or subtleties:

Earlier model: https://www.ebay.com/itm/IBM-00P4474-00P...SwN8FZrd6R

Later model: https://www.ebay.com/itm/IBM-80P7125-284...xyeZNTUZEB

The earlier model was sold with the RS/6000s, and the later model was sold with the IntelliStation POWER series. I'm going to wager a million bucks that the RS/6000 GXT4500Ps don't actually work with the IntelliStation POWER units, but it's too late as I already bought two "RS/6000" GXTs without realizing it. But we'll soon know for sure.

If this is the case then that really sucks and there are still three lingering questions I have:
#1 Is it possible to actually run two GXT6500Ps in a single system for multi-monitor support and ignore IBM's documentation
#2 Do two GXT4500ps even provide multi-monitor support at all despite getting this inferred from IBM's documentation.
#3 Maybe one could mix and match a GXT6500P and GXT4500P

Who knows. These are experiments I'll just have to do in time x_x

EDIT: Looks like IBM's documentation claims "00P4476" (the older RS/6000 style) is compatible with the IntelliStation POWER series:
https://www.ibm.com/support/knowledgecen...system.htm

2550 | POWER GXT4500P | graphics adapter | 7047-185, 9111-285, 9133-55A | 80P7124*
00P4476**

Not sure what to say... but we'll see.

[micrex22]

Well, both early rev. GXT4500Ps took to the system just fine. And as you can see--multi-monitor support does work.