Operating System - HP-UX
1752578 Members
4353 Online
108788 Solutions
New Discussion

Re: modern/current "processor architecture" information?

 
SOLVED
Go to solution
Doug O'Leary
Honored Contributor

modern/current "processor architecture" information?

Hey;

 

My client's trying to come up with a 'standardized' hardware information collection script across multiple operating systems and server types.  One of the bits of information they're looking for is 'processor architecture' and, for HPs, they're looking for this information in /usr/sam/lib/mo/sched.models.

 

For instance, my venerable TE520 on which I learned HPUX lo these many years ago, apparently has a processor architecture of PA7100.  

 

the sched.models file doesn't seem to have been updated in at least 5 years and doesn't contain the model strings for any of my ia64 systems.  

 

Does this type of query even make sense on modern systems (other than saying it's ia64)?  If so, is there an automated way of obtaining this information available?

 

Thanks for any hints/tips/suggestions.

 

Doug O'Leary

 

 


------
Senior UNIX Admin
O'Leary Computers Inc
linkedin: http://www.linkedin.com/dkoleary
Resume: http://www.olearycomputers.com/resume.html
5 REPLIES 5
Dennis Handly
Acclaimed Contributor

Re: modern/current "processor architecture" information?

>/usr/sam/lib/mo/sched.models.

 

The compilers stopped using a similar file when they could directly query the hardware and so stopped updating that file.

 

>the sched.models file doesn't seem to have been updated in at least 5 years

 

That's because PA2.0 was the last major change and the scheduling was the same.

 

And for Integrity, you can use machinfo(1).

And the compilers there just look at the CPUID register.

Doug O'Leary
Honored Contributor

Re: modern/current "processor architecture" information?

Hey;

 

Thanks for the response.  So, the ia64 architecture, from machinfo, is listed in the 'processor model' line?

[[snip]]

 

    process model:    2   Intel(R) Itanium 2 processor

 

[[snip]]

 

any idea how to get this info from a newer pa-risc.  machinfo doesn't exist, print-manifest doesn't provide it, and rp3440s don't show up in that antique file.

 

Thanks again for the reply.

 

Doug 


------
Senior UNIX Admin
O'Leary Computers Inc
linkedin: http://www.linkedin.com/dkoleary
Resume: http://www.olearycomputers.com/resume.html
Dennis Handly
Acclaimed Contributor

Re: modern/current "processor architecture" information?

>the ia64 architecture, from machinfo, is listed in the 'processor model' line?

>    processor model:    2   Intel(R) Itanium 2 processor

 

11.31 looks slightly different:

CPU info:
  2 Intel(R) Itanium 2 9100 series processors (1.67 GHz, 18 MB)
          666 MT/s bus, CPU version A1
          4 logical processors (2 per socket)

 

And you'll need to use -v to get more details.

 

>any idea how to get this info from a newer pa-risc.  machinfo doesn't exist

 

A newer PA on 11.31 has machinfo.  :-)

 

You can use getconf(1).

$ typeset -i16 x=$(getconf CPU_CHIP_TYPE); echo $x
16#20010104

 

But you would have to match the IDs to the chip names:

16#285 == 645 == PA 8800

Here is what "machinfo -v" shows for my rp3440:

CPU info:
  2 PA-RISC 8800 processors (1 GHz, 64 MB)
          CPU version 5
          4 logical processors (2 per socket)
          Vendor identification:        Hewlett-Packard
          Processor version info:       0x0000000000000285
          Processor capabilities:       0x00000000000000b2
                  FSTWS, FSTWX, FLDWS, FLDWX generate word tx to I/O space
                  Multimedia halfword shift-and-adds
                  Multimedia halfword adds and subtracts
                  Speculative data prefetch on loads of GR0
          Instruction cache:      64 MB
          Data cache:             64 MB

Bill Hassell
Honored Contributor
Solution

Re: modern/current "processor architecture" information?

Yes, sched.models is so last century...

 

SO the short answer is that you have to make a *lot* of tests to see what you have. There is very little in common between PA and IA, especially with machine architecture. The model command is found on version 10.20 through 11.31but it is just a marketing model number, not a specific processor architecture. Here's a few:

 

9000/800/L1500-6x
9000/800/N4000-44

ia64 hp server rx7640

9000/800/A180c
9000/800/rp7410
ia64 hp server rx2660

9000/800/rp7420

9000/800/SD32000

ia64 hp server rx4640

9000/800/rp4440

9000/800/L3000-8x

 

But these do reflect the architecture. The machinfo command is much better:

 

# machinfo
CPU info:
  1 Intel(R) Itanium 2 9100 series processor (1.67 GHz, 18 MB)
          666 MT/s bus, CPU version A1
          2 logical processors (2 per socket)

Memory: 8170 MB (7.98 GB)

Firmware info:
   Firmware revision:  04.11
   FP SWA driver revision: 1.18
   IPMI is supported on this system.
   BMC firmware revision: 5.25

Platform info:
   Model:                  "ia64 hp server rx2660"
   Machine ID number:      a7cfdfb6-8cbb-11dd-ab98-413d2c4b1b1b
   Machine serial number:  USE0808AEN

OS info:
   Nodename:  atl6
   Release:   HP-UX B.11.31
   Version:   U (unlimited-user license)
   Machine:   ia64
   ID Number: 2815418294
   vmunix _release_version:
@(#) $Revision: vmunix:    B.11.31_LR FLAVOR=perf

 

 

But this is only available on Itanium systems. So the most broadest information collection tool is probably cfg2html. It runs on version 10 through 11.31 and has a lot of if/else code to handle the wide range of hardware for HP-UX. And with versions that run on Linux, AIX and Solaris, it is a good fit for documenting multiple systems. The availability of both html as well as text documents makes scripting a lot easier.

 

www.cfg2html.com



Bill Hassell, sysadmin
Doug O'Leary
Honored Contributor

Re: modern/current "processor architecture" information?

Hey;

 

>>So the most broadest information collection tool is probably cfg2html.

 

Ya know, that is an absolutely brilliant idea.  I'm even familiar with cfg2html and, for some brain-damaged reason, I never thought of that.  I'll probably have some entertainment convincing the PHBs to do this; however, as everyone's always complaining about task saturation, maybe it won't be such a hard sell afterall.

 

Thanks again.  Absolutely brilliant.

 

Doug


------
Senior UNIX Admin
O'Leary Computers Inc
linkedin: http://www.linkedin.com/dkoleary
Resume: http://www.olearycomputers.com/resume.html