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 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> <1399.6866T1439T2714@pop.uky.edu> 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