±±±±±±±±±Ü ±±Ü ±±Ûßßß±±Û ±±Û ±±Û ±±Û ±±±±±±Ü ±±±±±Ü±±Ü ±±±±±±±±Ü ±±±±Ü ±±Ü ±±Ü ±±±±±±±±Ü ±±Û ±±±±±±±±Û ±±Ûßßßß ±±Û±±Û±±Û ßß±±Ûßßß ±±Ûß ±±Û ±±Û ±±Û±±Û±±Û ±±Û ±±Ûßßßßßß ±±±±Ü ±±Û±±Û±±Û ±±Û ±±Û ±±Û ±±Û ±±Û±±Û±±Û ±±Û ±±Û ±±Ûßß ±±Û±±Û±±Û ±±Û ±±Û ±±Ûܱ±Û ±±Û ßß±±Û ßß ±±±±Û ±±±±±±Ü ±±Û±±±±±Û ±±Û ±±±±Û ±±±±±±Û ±±Û ±±Û ±±Ü ßßßß ßßßßßß ßß ßßßßß ßß ßßßß ßßßßßß ßß ßß ßß Sat 12-10-1994 14:14:42 VERSION 2.1 Hi There... Well... I'm a little more than chagrined. I uploaded Version 1.0 of this little ditty just two days ago as I ASSumed all the of flack and chatter about the PENTIUM long division (floating point) errors was nothing more than the garden variety errors seen with computers and floating point mathematics. Part of my stupidity was and is... I have no access to a PENTIUM computer!!!! So I was and am really flying blind with this problem. So I uploaded this to get some feed back and to give you guys another tool to use on your big-time PENTIUM computers. Was nosing around CompuServe to see if I could add a routine that would allow me to load the CPU number (80286, 80386, 80486, 80586) into my QuickBASIC program. Was delighted to find one and added it to the executable file in this file set. Alas, it only goes up to "80486" so if you have a PENTIUM chip, it won't recognize it or will display "80486." PENTIUM2.EXE is therefore =NOT= an exact duplicate of PENTIUM.BAS PENTIUM2.EXE contains code that will NOT run inside the QBASIC.EXE interpreter that comes with DOS 5.0 (or newer). If you have version one (1), of this file set, you can see the difference between these two programs. PENTIUM2.EXE not only displays the "CPU" but also the "DOS VERSION" -and- throws up a red border -outside- your regular screen display area. After feeling flushed with success, I was about to upload this "improved version" to CompuServe and saw a number of file sets that showed multiple number pair errors with the PENTIUM chip; all using the same simple minded formula that I used. Worse... they could be reproduced by entering them into a spreadsheet in LOTUS or EXCEL!!! And... the PENTIUM errors caused DOLLARS & CENTS errors in their calculations! Wow! So, really need to have you PENTIUM cats get back to me, as this program -may- confirm this PENTIUM problem by showing a CUMULATIVE -sum- of ERRORS. This does NOT occur with a non-PENTIUM chip. Please get back to us poor people here in Santa Monica who just run 486/50s.... The next is my original text, just to show that I believe in the truth, the whole truth, and nothing but the truth. ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ VERSION 1.0 ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ Trying to understand what all the hoopla in the newspapers and on Internet is over discovering that the Pentium chip.... like all CPUs... makes a "floating point" math error. If you do long division math (using numbers with decimal places) with a computers, you "always" get some "rounding errors." I think that is true for all computers, not sure but think so. Apparently the Internet community is up in arms because they don't know, that. Or... else, everyone on the facility of State U. just wants Intel to buy them a new computer! The formula given in the New York Times November 24th, 1994 was: ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ 4,195,835 - [(4,195,835 / 3,145,727)*3,145,727] = Answer   (First Number) (Second Number) ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ The correct "Answer" is Zero. But with this formula and a Pentium chip, the answer is "256" -or- an error of 256/4,195,835 or 61/100,000. A teeny tiny error unless you are looking to set a new world's record in "prime numbers" as the math person in New York was trying to do when he discovered this little Pentium chip crack. You get a zero (0); because that formula divides the First Number by the Second Number to get a fraction (a long division or floating point number) then multiplies that result by the Second Number. It should be clear to anyone but the hopelessly lost that a number divided by a second number and then multiplied by that second number is always the FIRST NUMBER. Then if you subtract the result from the first number it is ZERO. But with computers that last statement just ain't necessarily so.... So.... I have taken it upon myself to give you a little program that will show you that your very own beloved computer of whatever make and kind will do just what the Intel Pentium chip does.... give you long division errors. If you have a Pentium computer and a less prestigious one, run this test on both of them. Get back to me if you find big differences in the amount of errors. Unlike all of my other stuff... this program is =both= compiled into a stand-alone *.EXE file and as a "RUNable" BASIC program that will work with QBASIC.EXE. I dearly hope it WENDs it way via ASCII type E-Mail to Internet so that those non-mathematical, non-computer types can run it and maybe add some light to the flaming and heat on Internet. Use the batch file RUN.BAT as it will start your QBASIC.EXE program (comes with DOS 5 and newer) and run PENTIUM.BAS. Get back to me.... I'm really curious to find out if the Pentium is really worser than the 286, 386, 486 CPUs. Again, to run PENTIUM.BAS program with QBASIC.EXE, just open it in that program and "RUN" it or use RUN.BAT which came with this file set. Phew! Just got back here.... from a long editing session with PENTIUM.BAS. Heck, I tried to make this a simple little program and it has grown like crazy as I tried to trap errors and put in a timer. So... if you find a "bug," get back to me.... Uh.... the timer is a CUMULATIVE time timer, so if you run the program more than once, it sums the times of all the runs... What can I say, except it works speedily that way. John De Palma on CompuServe 76076,571