We've been involved with a couple comparisons and here's what we've
discovered (FWIW, this is not very scientific):


HPUX is slow compared to MPE
HPUX does not multi-task very well.  You need a seperate system for each
application (and I've had some HPUX gurus tell me that, too).  That's
partly why I see multiple 9000s in a datacenter versus a single 3000 in others

As a result of the above, a 100 Mhz 3000 may outperform a 440 Mhz 9000 if
both are running multiple applications

And I repeat, this was not scientifically documented...my sample space is 2!

John Lee
Vaske Computer Solutions




At 08:50 AM 3/13/03 -0600, Mark Landin wrote:
> On Wed, 12 Mar 2003 11:16:12 -0600 (Central Standard Time), Ed Archer
><[log in to unmask]> wrote:
>
>>If I remember correctly, a while back, someone posted
>>performance comparisons between various 3000 MPE
>>systems and 9000 HPUX systems.
>>I am specifically interested in the relative
>>performance comparison between the 9X9 MPE servers,
>>and the N and L Class HPUX servers.
>>Any help will be appreciated.
>
>
>Since there are no common "benchmarks" between the two platforms,
>there is really no *meaningful* peformance comparison bewteen the two
>platforms. By meaningful, I mean one which would help you predict with
>a good level of confidence that the HP-UX system you bought would do
>as much work as your MPE system.
>
>Frankly, the hardware on the 9000 side is much more powerful than on
>the 3000 side. Someone at HP decided to throttle the A- and N- class
>MPE boxes to 1/3 of their real clock speeds seen in their 9000
>counterparts. Further, the 9000 has a much higher "top end" (see:
>Superdome) than the 3000s. Conversely, MPE seems to get more work done
>with a given set of hardware resources than HP-UX does. There are many
>variables here besides just CPU speed and I/O backplane.
>
>If you are trying to size an HP-UX system to replace your 3000 system,
>the only way to really do it is to rent a 9000, install your stuff on
>it, and benchmark it. If it's too small, rent a bigger one and try
>again.
>
>Then let us know what you discovered. :)
>
>* To join/leave the list, search archives, change list settings, *
>* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
>
>

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *