| Author |
OpenVMS and non-HP Itanium systems |
ch_123
Member
Posts: 6
Joined: 08.06.10 |
| Posted on June 08 2010 04:41 |
|
|
|
Will the IA64 port of OpenVMS only run on Integrity servers and HP workstations? I have an SGI Altix system which I'd be interested in running OpenVMS if the IA64 hobby kit ever comes out. Is this possible? |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
imiller
Administrator
Posts: 277
Location: UK
Joined: 24.02.06 |
| Posted on June 09 2010 09:29 |
|
|
perhaps. Initial porting work for VMS I64 was done on some Itainium boxes from Intel.
There is nothing intentional to prevent VMS being run on other Itainium systems but nothing done to make it work either. |
 |
| Author |
RE: OpenVMS and non-HP Itanium systems |
ch_123
Member
Posts: 6
Joined: 08.06.10 |
| Posted on June 13 2010 13:45 |
|
|
I was thinking that that might be the case...
So, does anyone have any IA64 media they're willing to share? =P |
|
|
| Author |
RE: OpenVMS on Altix |
saq
Member
Posts: 60
Joined: 24.05.08 |
| Posted on July 07 2010 12:35 |
|
|
OpenVMS will not work on Altix unless HP specifically codes support in. SGI Altix is based on the old Origin ccNUMA architecture, which is very different from Integrity's setup.
Provided that you can get a generic Itanium with supported chipset you should be able to get it to work, but Altix is too specialized. It scales to substantially more processors than any HP offering (SGI supports 512 sockets in Altix, HP only offers 32 in Superdome). |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
eMGee
Member
Posts: 87
Location: The Netherlands
Joined: 28.03.10 |
| Posted on January 02 2011 16:12 |
|
|
SAQ, have you tried it? I don't see how ccNUMA, which operates on a different level from the EFI (I should imagine, like with SGI's MIPS/IRIX systems) and thus wouldn't conflict in theory, I'd imagine.
(Speaking of NUMA, I believe I recently saw a reference to NUMA and OpenVMS. Although I'm not sure if it's the same thing.)
Edited by eMGee on January 02 2011 16:15 |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
imiller
Administrator
Posts: 277
Location: UK
Joined: 24.02.06 |
| Posted on January 05 2011 04:00 |
|
|
|
V8.4 has support for the NUMA characteristics of the new BL890 i2 blades |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
eMGee
Member
Posts: 87
Location: The Netherlands
Joined: 28.03.10 |
| Posted on January 06 2011 11:18 |
|
|
Nice, so I didn't misread that!  |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
ch_123
Member
Posts: 6
Joined: 08.06.10 |
| Posted on February 25 2011 13:25 |
|
|
|
saq wrote:
OpenVMS will not work on Altix unless HP specifically codes support in. SGI Altix is based on the old Origin ccNUMA architecture, which is very different from Integrity's setup.
Provided that you can get a generic Itanium with supported chipset you should be able to get it to work, but Altix is too specialized. It scales to substantially more processors than any HP offering (SGI supports 512 sockets in Altix, HP only offers 32 in Superdome).
The machine I have only has 4 CPUs though. Unless of course NUMA compatibility proves to be an issue. |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
eMGee
Member
Posts: 87
Location: The Netherlands
Joined: 28.03.10 |
| Posted on March 03 2011 10:54 |
|
|
It wouldn't hurt to give it a try, right? (Keep in mind, as you may know, with VMS you can't multi-boot from the same disk. So, it would probably be best if you used a separate disk.)
Edited by eMGee on May 10 2011 02:44 |
|
|
| Author |
RE: OpenVMS on Altix |
saq
Member
Posts: 60
Joined: 24.05.08 |
| Posted on May 06 2011 19:58 |
|
|
You can give it a try, but I certainly wouldn't spend (or bet) any money on it.
SGI's NUMA ASICs are proprietary and different from anything that HP offers, as are the other parts of the SGI chipset. Integrity OpenVMS machines use HP chipsets, and while I am far from an expert on Itanitum machines I haven't seen them used in any other manufacturer's Itantium systems.
If you look at the Linux source you can see the parts that are specifically written for Altix support, and that they're different from the Integrity devices. For instance, many Altixes use the IOC4 I/O ASIC as used on Origin 3x machines (not sure about the newer Altii), and the kernel source has the following comments:
The SGI IOC4 PCI device is a bit of a strange beast, so some notes on
it are in order.
First, even though the IOC4 performs multiple functions, such as an
IDE controller, a serial controller, a PS/2 keyboard/mouse controller,
and an external interrupt mechanism, it's not implemented as a
multifunction device. The consequence of this from a software
standpoint is that all these functions share a single IRQ, and
they can't all register to own the same PCI device ID. To make
matters a bit worse, some of the register blocks (and even registers
themselves) present in IOC4 are mixed-purpose between these several
functions, meaning that there's no clear "owning" device driver.
The solution is to organize the IOC4 driver into several independent
drivers, "ioc4", "sgiioc4", and "ioc4_serial". Note that there is no
PS/2 controller driver as this functionality has never been wired up
on a shipping IO card.
There also appear to be specialized code for the SHub (HUB/BEDROCK chip) and the XIO I/O system that HP would have no reason to put in VMS.
Based on this I hold out very low hopes for running OpenVMS on it. Pity, as the Altix is a nice architecture, and OpenVMS is a great OS, but I fear that never the twain shall meet.
Edited by saq on May 06 2011 19:59 |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
eMGee
Member
Posts: 87
Location: The Netherlands
Joined: 28.03.10 |
| Posted on May 10 2011 02:40 |
|
|
|
saq wrote:
You can give it a try, but I certainly wouldn't spend (or bet) any money on it.
[...]
Based on this I hold out very low hopes for running OpenVMS on it.
Rest assured, nobody in the thread so far has given anyone any false hopes nor expressed any intention of trying. If anyone who already owns SGI IA-64 (i.e. Altix) hardware ― with a spare disk and an VMS I64 installation set ― wants to try it, then it's obviously not at all expensive.
SGI's NUMA ASICs are proprietary and different from anything that HP offers [...]
There also appear to be specialized code for the SHub (HUB/BEDROCK chip) and the XIO I/O system that HP would have no reason to put in VMS.
Has anyone claimed otherwise in this thread or suggested anything contrary to the above? Also, why would someone need XIO under VMS?
Edited by eMGee on July 21 2012 01:43 |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
saq
Member
Posts: 60
Joined: 24.05.08 |
| Posted on June 17 2011 05:55 |
|
|
|
All I/O on IA64 Altix and recent MIPS boxes (except O2) goes out over XIO initially and then is bridged to PCI by either an XBRIDGE or BRIDGE chip. If your OS doesn't support XIO then you don't have any I/O you can get at. |
|
|
| Author |
RE: I see you have difficulty reading, or perhaps you felt the urge to give us a sermon (as always). |
eMGee
Member
Posts: 87
Location: The Netherlands
Joined: 28.03.10 |
| Posted on June 18 2011 10:17 |
|
|
Does the bridge rely on any higher level software to function? Either way, what I meant (in my previous post) was the necessity of XIO expansion slots, not so much the bus.
Again, nobody expects it to work. Anyone who wants to seriously use VMS I64 will buy an HP IA-64 (Integrity) system instead. As I said, only if you just so happen to already own a SGI IA-64 system it'd be worth the try doing anything with it (with the risk of likely disappointment).
I hope to have hereby clarified this matter for you. My apologies if I wasn't more clear before, I hope I'm now.
Edited by eMGee on July 21 2012 01:42 |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
ch_123
Member
Posts: 6
Joined: 08.06.10 |
| Posted on September 16 2011 03:36 |
|
|
Managed to get a copy of an OpenVMS I64 installer disk, tried booting from it on the Altix, got the following error -
[QUOTE]*** smbios_SystemSerialNumber error
*** smbios_PhysicalSSN error
ERROR: Unable to release secondary CPUs - Invalid PIB address.
WARNING: Unable to release secondary CPUs.
%VMS_LOADER-W-Unable to identify primary console device.
> Select a primary console from the Boot Configuration Menu.
> Attempting to continue with a default console device.
[/QUOTE]
Well, I guess it checks that it's running on HP hardware. It also seems to be depending on some HP-specific EFI stuff. Was worth a try anyway  |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
eMGee
Member
Posts: 87
Location: The Netherlands
Joined: 28.03.10 |
| Posted on July 03 2012 07:41 |
|
|
ch_123 wrote:
Managed to get a copy of an OpenVMS I64 installer disk, tried booting from it on the Altix, got the following error -
*** smbios_SystemSerialNumber error
*** smbios_PhysicalSSN error
ERROR: Unable to release secondary CPUs - Invalid PIB address.
WARNING: Unable to release secondary CPUs.
%VMS_LOADER-W-Unable to identify primary console device.
> Select a primary console from the Boot Configuration Menu.
> Attempting to continue with a default console device.
Interesting, thanks for sharing (and trying), I must've missed this post at the time.
Yes, it looks like these are fundamental chipset/firmware errors, at a first glance. I think you were the first person to ever try and definitely confirm that it doesn't work, because the only references to the above error messages ― in major web search engines ― lead right to this forum thread and your post in particular.
What I find find very interesting, is that VMS does load to some degree. It boots enough to generate a VMS error code, despite the errors.
What you could try, is to absolutely and positively make sure that you configured just one output and to the [serial] “console†(or whatever it's called in the SGI EFI environment). I mean, the error is also common on some HP Integrity systems, when multiple outputs are configured or when the output is misdirected.
Well, I guess it checks that it's running on HP hardware. It also seems to be depending on some HP-specific EFI stuff. Was worth a try anyway
I have no idea what the messages exactly mean, although “smbios_SystemSerialNumber†is quite self-explanatory, but it's still pretty interesting and I thank you for your try. (If you want to try what I mentioned in my previous paragraph, that would be great; if not, that's also fine.)
Edited by eMGee on July 09 2012 00:03 |
|
|
| Author |
RE: OpenVMS and non-HP Itanium systems |
ch_123
Member
Posts: 6
Joined: 08.06.10 |
| Posted on July 20 2012 12:52 |
|
|
Around the same time as I tried to boot VMS, I also tried firing up a copy of HP-UX IA64. It appeared to start booting successfully, but then it would stop feeding information to the console. A somewhat similar response, except there were no obvious errors.
The likely cause of this problem is that on the Altix 350, the serial port is part of the, to use the SGI nomenclature, "L1 controller." This device gives you the ability to power up/down and run diagnostics on the machine, and others attached via NUMA cables, even when they are powered down. Because it's not a regular old serial port, it has a different device identifier to a regular serial port (under Linux, for example, it is /dev/ttySG0 as opposed to /dev/ttyS0.) When setting up Linux on the Altix, you specify that you want to use ttySG0 by means of a kernel argument when launching the installer image from the EFI shell.
It seems for HP UX and VMS/IA64 that this is done through a setup screen. The Altix does not have an equivalent to this though. I get the impression that the HP setup screen is some sort of proprietary voodoo that HP's OSes expect, but is absent on other vendor's machines, but I may be wrong here.
That said, I wonder if I can find anything in the Altix's documentation about this (there are some configuration options in the firmware, but I don't recall there being any to do with serial output), or if perhaps some sort of PCI serial card would act as the ttyS0 device (or the appropriate equivalent under HP UX or VMS/IA64.)
Edited by ch_123 on July 20 2012 12:53 |
|
|
