HP3000-L Archives

February 2000, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jerry Fochtman <[log in to unmask]>
Reply To:
Jerry Fochtman <[log in to unmask]>
Date:
Tue, 8 Feb 2000 08:54:21 -0600
Content-Type:
text/plain
Parts/Attachments:
text/plain (63 lines)
At 06:31 PM 2/7/2000 -0500, Richard Gambrell wrote:
> >Also, 6.5 is said to make better use of more processors (perhaps improves
> >performance on all? multiple processor configurations). Dispatcher and
> >scheduler were said to have been rewritten to make this possible by
> >reducing the single threading bottlenecks.


Although I'm simply trying to illustrate our performance experiences with the
new hardware configurations & the MPE/iX 6.5 release, some might consider the
following a...

<plug>

We have a some bench mark tests that we use to track changes in DBGeneral's
performance as we enhance the tool.  One of these tests involves conducting
a structural change on a 65,000,000 entry detail set.  Almost 2 years ago
this test took approximately 54 minutes to run on a 996/800 system under 5.5
with the maximum memory configuration. We then applied some significant design
changes in DBG and ran this same test again on 5.5 using a 996/600 with the
same amount of memory and essentially the same disc environment.  This time
the test completed in 26 minutes.

At HP's request, this fall we ran the same test using 6.5 on the new 998/1200
(or was it a 989/1200?) system with 16GB of memory, again using the exact same
data volume.  In this case HP has made a number of changes to the operating
system
and we have incorporated additional performance enhancements in DBG.  The net
result is our same test completed in just over 5 minutes!  Truly an amazing
level of performance improvement which will benefit current and future
HP3000 users!

Granted these are not very scientific bench mark comparisons, as a lot of
different factors have changed between tests.  However, it does show just
where things have come from over the last couple of years in both hardware
and software design and integration. Also, it looks like I'm going to have
to get some larger test volume (in the 100+million entry range) so DBG will
execute longer and we can collect better performance characteristics... :)

As with all bench mark data, the standard disclaimers apply...(YMMV!).

Way to go HP/CSY!

</plug>


/jf
                               _\\///_
                              (' o-o ')
___________________________ooOo_( )_OOoo____________________________________

                          Tuesday, February 8th
                     U.S. Boy Scouts founded in 1910
         (Also, Joseph's Pinewood car won 1st place in the Pack!)

           Today in 1693 - College of William and Mary, 2nd is U.S.,
                           was chartered.

___________________________________Oooo_____________________________________
                             oooO  (    )
                            (    )  )  /
                             \  (   (_/
                              \_)

ATOM RSS1 RSS2