Re: 10.20 vs 11.00 Performance comparisons (84 Views)
Reply
Honored Contributor
Alan Riggs
Posts: 881
Registered: ‎03-08-1998
Message 1 of 7 (84 Views)

10.20 vs 11.00 Performance comparisons

Hello, I am investigating the possibility of system upgrades from 10.20 to 11.00 32-bit. I recall seeing at one point a FAQ or white paper which mentioned possible performance degradation on teh 11.00 platform. My own searches through the docs produced this FAQ:

http://docs.hp.com/hpux/content/osfaq.html

which is close, but I recall a slightly more detailed treatment. Can anyone point me to a fuller treatment of the performance impact of an 11.00 32-bit upgrade and/or share a personal experience of same?
Please use plain text.
Respected Contributor
Robert Gamble
Posts: 419
Registered: ‎04-11-1997
Message 2 of 7 (84 Views)

Re: 10.20 vs 11.00 Performance comparisons

Alan,

Consider what applications you will be using. If the application is tuned for 64 bit, then use the 64 bit HP-UX.
Running a 32 bit app on a 64 bit OS _will_ cause a performance degradation.

I personally have seen HP-UX 11 run slightly slower on older 800 & 700 series machines, but unable to see difference on newer systems, like a K580 or D390. N-Class of course only run 64 bit 11.

Hope this helps!
Please use plain text.
Honored Contributor
Alan Riggs
Posts: 881
Registered: ‎03-08-1998
Message 3 of 7 (84 Views)

Re: 10.20 vs 11.00 Performance comparisons

Thanks Robert, but the situation I am looking at involves machines with less than 4GB of RAM and an application that will not be recoded for 64-bit compilation. Given that, it seems almost guaranteed that we will suffer a performance degredation if going to a 64-bit version of 11.00.
Please use plain text.
Frequent Advisor
Rumen Ginev
Posts: 47
Registered: ‎05-07-2000
Message 4 of 7 (84 Views)

Re: 10.20 vs 11.00 Performance comparisons

Hi Alan!

We keep in our company an extra HP9000/847/G30 mashine for experiments. I've installed on it both 10.20 and 11.00 for the same purpose - performance comparison. Find attached to this massage the output from /opt/ignite/bin/print_manifest for detailed description of the system.
There is obvious performance degradation in running the same applications on 10.20 and 11.00. You can see this even when use some basic HP-UX utilities such sam and swinstall.
I don't have benchmarks to send, but if you want further details or you can suggest experiments to carry out just tell me
Please use plain text.
Honored Contributor
Andy Monks
Posts: 462
Registered: ‎05-31-1999
Message 5 of 7 (84 Views)

Re: 10.20 vs 11.00 Performance comparisons

While initially, hp-ux 11.0 was slower than 10.20, lots has happened since it first came out. When the V2500 the kernel had to perform at a required level. This lead to profile based optimization of the kernel.

While it still maybe slower at some things (sam, sd-ux), the overall system calls are very fast (with the latest patches).

The only way to be sure if it's fast enough for you, is to benchmark it.

Hope that helps.
Please use plain text.
Honored Contributor
Alan Riggs
Posts: 881
Registered: ‎03-08-1998
Message 6 of 7 (84 Views)

Re: 10.20 vs 11.00 Performance comparisons

Thanks for the info. Roumen, I cannot think of specific, detailed benchmarks which would translate from your system to mine. But if you could perhaps send me some simple timex outputs for system commands and perhaps some scripts which exercise I/O. I would appreciate any such information. You can email me at:

riggsa@sensormatic.com
Please use plain text.
Honored Contributor
CHRIS ANORUO
Posts: 1,361
Registered: ‎05-29-1998
Message 7 of 7 (84 Views)

Re: 10.20 vs 11.00 Performance comparisons

Hi Alan,

Try the attached Kernel parameters from a 10.20 9000/800/K580 server. It is the current kernel that I have for my oracle 8.0.5 applications and we don't have performance problems. Apply the parameters applicable to your systems's kernel (11.00).


Cheers!
When We Seek To Discover The Best In Others, We Somehow Bring Out The Best In Ourselves.
Please use plain text.
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation