From - Sun Oct 27 22:14:42 1996
Path: 
nntp.ulg.ac.be!news.belnet.be!swsbe6.switch.ch!swidir.switch.ch!surfnet.nl!howla
nd.erols.net!news-
peer.gsl.net!news.gsl.net!hammer.uoregon.edu!news.uoregon.edu!newsadmin
From: Steve VanDevender <stevev@efn.org>
Newsgroups: comp.sys.hp48
Subject: Re: HP 48 Speed
Date: 24 Oct 1996 23:59:50 -0700
Organization: University of Oregon
Lines: 203
Sender: stevev@tzadkiel.efn.org
Message-ID: <874tjjcyvt.fsf@tzadkiel.efn.org>
References: <546ccf$otp@carabelli.com> <32673002.41C67EA6@matstat.unit.no>
	<Pine.SUN.3.91.961019173236.27322B-100000@cnct.com>
	<1399.6866T1439T2714@pop.uky.edu>
	<Pine.SUN.3.91.961020143122.17673C-100000@cnct.com>
NNTP-Posting-Host: cisco-ts12-line1.uoregon.edu
NNTP-Posting-User: stevev
X-Newsreader: Gnus v5.2.25/XEmacs 19.14

jose@cnct.com writes:
> > My clock is a little screwy now..  My theory is that the electrons are 
moving
> > faster than the normal speed of sound, and actually approching the speed of
> > light, creating a small wake in time and making my clock run a bit wrong.
> 
> Have you tried using the IR port yet? Maybe you can project some of these 
> speeding electrons and form a time/space warp? So _THAT'S_ how they do it 
> on sliders!
> 
> 	Jose

No, actually it's even easier than that.  During my previous career as
Chief Research Scientist at Speed Merchant Laboratories, I reported on
some anomalous HP 48 behavior we discovered while doing research on
battery life, which I include below:

In-reply-to: an1060@anon.penet.fi's message of Sat, 27 Feb 1993 18:32:02 GMT
Newsgroups: comp.sys.hp48
Subject: Re: Memory & Power Consumption
References: <1993Feb27.184413.12699@fuug.fi>
Distribution: 
--text follows this line--
In article <1993Feb27.184413.12699@fuug.fi> an1060@anon.penet.fi writes:

   Hello,

   I hope this is not an FAQ...But...

   I would like to know if the HP-48's power drain on the batteries is 
   significantly increased by more programs/data stored in RAM...
   If so, then can someone quantify this relation??

   Thanks in advance!!!

   BTW:  I am also interested in how this applies to RAM cards...

We're glad you asked that, Mr. Mantell, ahem, we mean Mr. an1060!
As the chief research scientist at Speed Merchant Laboratories (a
wholly owned subsidiary of Speed Merchant, Inc.) I have been
studying this question for some time.

We at Speed Merchant use a variety of Hewlett-Packard equipment,
and HP 48SX calculators are standard issue to our research and
engineering staff.  Corporate pennypinching has forced us to
ration batteries to our staff, and therefore finding ways to
extend battery life has become an issue of some concern.  We have
run a series of experiments on this matter, and here present our
findings:

HP 48SX calculators were loaded with special memory patterns,
ranging from all zeros to all ones, leaving some available memory
to allow calculator operation, on calculators with no expansion
memory, one 128K memory card, or two 128K memory cards.  The time
from installation of fresh batteries (we used Kodak Supralife
brand) until the low battery light came on was then recorded; due
to differing rates of use this time shows some variation which
was quite consistent.

                    HP 48SX with
                32K       160K       288K
% of binary
1s in memory

  0%         2160 hrs   2160 hrs   2160 hrs  (+/- 720 hrs)

 25%          90 days    90 days    90 days  (+/- 30 days)

 50%         12 weeks   12 weeks   12 weeks  (+/- 4 weeks)

 75%         3 months   3 months   3 months  (+/- 1 month)

100%       0.25 years 0.25 years 0.25 years  (+/- 0.83 years)

As you can see from the rapidly decreasing numbers, increasing
proportions of binary 1s in calculator memory make a tremendous
difference in battery life.  In general the 25-50% figures
represent typical usage, where memory is kept full or nearly full
assuming that stored data has a distribution of roughly half ones
and half zeros when expressed in binary form.

As a result of this research we are now working on ways to reduce
the number of binary 1s stored in HP 48 calculator memory using
various data compression techniques, but these will not be on the
market for some time.  Until then, we can suggest a few methods
for reducing the number of 1s stored in your calculator's memory:

1.  If a GROB contains more dark pixels that light pixels, use
the NEG command on it before storing it in memory, and use NEG
again just before placing it into the display.  NEG on GROBs will
change dark pixels to light and light to dark; when used on a
mostly-dark GROB this will then reduce the number of 1 bits
needed to store it.

2.  Use as few digits of precision as are needed in the problem
at hand, and use the RND or TRUNC commands before storing results
to truncate the excess digits and remove unneeded 1 bits.

3.  Turn off unneeded display annunciators; besides requiring 1
bits in memory, activated annunciators also draw more power in
the display.

4.  Clear the PICT and stack whenever the data in them is
unneeded.

5.  Garbage collect frequently; don't wait until the heap is
exhausted.  Make it a habit to do MEM DROP every few keystrokes.

We at Speed Merchant hope that this information has proven
useful.  Once again, look for our soon-to-be announced HP 48
software package that will automatically minimize the number of 1
bits in memory and thereby improve your battery life in typical
usage from as little as 0.17 years to as much as 1440 hours!

Steve VanDevender
Chief Research Scientist
Speed Merchant Laboratories
a wholly owned subsidiary of Speed Merchant, Inc., manufacturers
of the Speed Merchant 5000 -- the only wheelchair with a roll
bar!

   -------------------------------------------------------------------------
   To find out more about the anon service, send mail to help@anon.penet.fi.
   Due to the double-blind system, any replies to this message will be 
anonymized,
   and an anonymous id will be allocated automatically. You have been warned.
   Please report any problems, inappropriate use etc. to admin@anon.penet.fi.
   *IMPORTANT server security update*, mail to update@anon.penet.fi for details.

Note that although you posted this anonymously, we have attempted
to keep your identity secret, Mr. Mantell, I mean, Mr. an1060.

In-reply-to: rrd@hpfcso.FC.HP.COM's message of 1 Mar 93 16:30:11 GMT
Newsgroups: comp.sys.hp48
Subject: Re: Memory & Power Consumption
References: <1993Feb27.184413.12699@fuug.fi> <107200052@hpfcso.FC.HP.COM>
Distribution: 
--text follows this line--
In article <107200052@hpfcso.FC.HP.COM> rrd@hpfcso.FC.HP.COM (Ray Depew)
writes:

   Mr. VanDevender graciously provided us with battery usage numbers, but
   he omitted the weight of all those binary 1's.  (You know, the more 1's
   in memory, the heavier the calculator weighs...)  Perhaps Speed Merchant can
   follow-up with a report on the comparative weights of different brands of
   battery, and a recommendation on the optimum battery to use for
   longest life, fullest RAM and lightest 48SX.  Now *there's* an engineering
   problem for you!

Actually, Speed Merchant Labs has been concentrating on another
research project of possibly staggering implications.

Some may have thought that one line of the published results on
battery life was in error.  It said that HP 48s whose memory had
been filled completely with binary 1s had an average battery life
of 0.25 years with a variance of 0.83 years, which is an order of
10 greater than the variance under other conditions, and which
would also imply that in some cases the batteries ran out several
months before they were installed.

We, of course, didn't believe these figures when the researcher
assigned to that part of the investigation presented them.  She,
however, had simply adhered to the measurement standards we had
set up for the experiment, and measured the time between battery
installation and appearance of a low-battery condition for
calculators given simulated normal daily use.  A few convincing
demonstrations by her showed that her results were not in error.

Her demonstrations were stunning.  After filling a calculator's
memory with all 1s, the calculator disappeared in a blue flash.
She did this with several calculators, then we proceeded to
search our building and grounds, and located about half of them,
many of which had appeared to have been lying around for months.
Several months later, one at a time, the other calculators were
recovered, in the same condition they had been in after leaving
her office.

Although we were reluctant to believe such an extraordinary
hypothesis, examining the affected calculators and their clocks
led us to an inescapable conclusion: filling HP 48 calculator
memory with all binary 1s causes the calculator to travel in
time.  Once Speed Merchant Labs has been able to calibrate and
quantify this effect, we expect that software exploiting it will
provide the HP 48 community with applications far more useful
than remote controls or tricorders.

Our only theory as to why this effect occurs is that a
combination of the high bit densities of modern memories,
combined with the small size of the HP 48, allows the formation
of a spacetime singularity which is capable of transporting the
HP 48 through space and time.  This effect does not happen with
larger devices because of their greater mass, although one of our
theoretical physicists is claiming that the effect will start to
appear in laptops when they begin using 16M-bit memory chips.

Steve VanDevender
Chief Research Scientist
Speed Merchant Laboratories
a wholly-owned subsidiary of Speed Merchant, Inc.


-- 
Steve VanDevender         USENET haiku: Take it somewhere else
email: stevev@hexadecimal.uoregon.edu   This thread does not belong here
WWW: http://jcomm.uoregon.edu/~stevev   Death to crossposters
