Re: Odd/Even packed numbers. -- MIDRANGE-L

Hi Al,

actually I have this RIDICULOUS problem:

I'm working in a Bank and not in a barber shop, and I've been asked to give
a technical answer to this question (and to decide). Maybe if I respond that
this is a RIDICULOUS question, in the future I will probably have to ask the
barber shop to give me a job.

At present a certain field is 7.0 - and it must be enlarged because now -
after 6 (six) years - it is going to become full.

How much we should enlarge it? With 8.0 we can add more 90,000,000 items and
it is enough for the next 50 years but it seems that there is an overhead
(this field is part of  a key in other db for a total of  800,000,000 rcds).
With 9.0 there is no overhead and no waste of space but we have hundreds of
panels (especially subfile lines) and reports where one more byte cannot
fit.

I ran a simple test (on a 830) adding 100,000,000 of records with 8.0 vs.
9.0 on a empty table with 2 of these fields (without keys or logicals) and
the overhead is 0 (ZERO) but +12% compared to 7.0 (however still nothing if
compared to the eternity)
These ase the times:
7.0 needs 66secs of CPU while 8.0 needs 74s and 9.0 always needs 74s.

I repeated the test with 2 logicals (2 keys - K1+K2 and K2+K1 on the same
7/8/9.0 pf); adding only 10,000,000 records gave these results:
7.0 needs 403secs of CPU while 8.0 needs 401secs and 9.0 needs 402secs, in
other words 7,8 or 9 makes no difference.

Because I remember that at s/38 times the semi-byte (oh sorry the NIBBLE)
did really matter, I wanted to get some other opinions.

I posted the same into MI400 newsgroup and I've been pleased to find there a
very interesting answer to this problem. Take a look, Al.

-- Giuseppe




----- Original Message -----
From: "Al Barsa" <barsa@barsaconsulting.com>
To: <midrange-l@midrange.com>
Sent: Friday, November 29, 2002 4:32 AM
Subject: Re: Odd/Even packed numbers.


>
> Happy Thanksgiving, you all.
>
> I think that in this day and age, with the CPU power that we take/squander

> for granted, this entire discussion is ridiculous.  In the early /38 days,
> we set counters in binary, because it made a difference.  Today, I think
> that the right thing is to set your data as you need it (from this
> perspective), and not waste time thinking about system efficiencies.  The
> brute power of the systems tat we use today far overweighs setting the
> length or type of fields.
>
> Al
>
> Al Barsa, Jr.
> Barsa Consulting Group, LLC
>
> 400>390
>
> 914-251-1234
> 914-251-9406 fax
>
> http://www.barsaconsulting.com
> http://www.taatool.com
>
>
>
>
>
>
>                       Glenn Ericson
>                       <Glenn-Ericson@att        To:
midrange-l@midrange.com, <midrange-l@midrange.com>
>                       .net>                     cc:
>                       Sent by:                  Subject:  Re: Odd/Even
packed numbers.
>                       midrange-l-admin@m
>                       idrange.com
>
>
>                       11/28/2002 09:58
>                       PM
>                       Please respond to
>                       midrange-l
>
>
>
>
>
>
> --
> [ Picked text/plain from multipart/alternative ]
> At 09:09 PM 11/28/2002 -0500, Booth Martin wrote:
>
> >(zoned & signed goes back to the days of punch cards.  A column punch was
> >zoned" if the top three rows had a punch. The top three rows were the
zone
> >rows - row 1 was A-J, row 2 was K-S, and a row 3 punch was the rest of
the
> >characterset.  If I am a bit off please forgive me  -  that was 30+ years
> >ago.)               Row 1  A-I (Plus  values)  Row 2 J-R (negative
> >values)  Row 3 Q-Z ( I don't recall)
> >
> >Leif, I believe your comment that signed fields take much longer, but 10
> >times longer?  oh gosh...  That is huge.  Did you save the metrics?
> >
> >
> >---------------------------------------------------------
> >Booth Martin   http://www.MartinVT.com
> >Booth@MartinVT.com
> >---------------------------------------------------------
> >
> >-------Original Message-------
> >
> >From: midrange-l@midrange.com
> >Date: Thursday, November 28, 2002 08:20:03 PM
> >To: midrange-l@midrange.com
> >Subject: Re: Odd/Even packed numbers.
> >
> >From: Leif Svalgaard <leif@leif.org>
> > > > > > From: Booth Martin
> > > > > > > Leif, would you try one more thing since you are already
there,
> > > please?
> > > > > > > What happens if you define it as 14S 0 and 15S 0? Does packed
> vs.
> >
> >yes, signed (aka zoned) is 10 times slower than packed.
> >_______________________________________________
> >This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
> list
> >To post a message email: MIDRANGE-L@midrange.com
> >To subscribe, unsubscribe, or change list options,
> >visit: http://lists.midrange.com/cgi-bin/listinfo/midrange-l
> >or email: MIDRANGE-L-request@midrange.com
> >Before posting, please take a moment to review the archives
> >at http://archive.midrange.com/midrange-l.
> --
>
> _______________________________________________
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
list
> To post a message email: MIDRANGE-L@midrange.com
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/cgi-bin/listinfo/midrange-l
> or email: MIDRANGE-L-request@midrange.com
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.
>
>
>
>
>
>
> _______________________________________________
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
list
> To post a message email: MIDRANGE-L@midrange.com
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/cgi-bin/listinfo/midrange-l
> or email: MIDRANGE-L-request@midrange.com
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.
>
>