UMFORTH@WEIZMANN.BITNET (01/24/86)
Date: Wed, 22 Jan 86 11:41 EST
From: "Maj. Douglas Hardie" <Hardie@DOCKMASTER.ARPA>
Subject: Re: N-Digit Pi Code
To: UMFORTH%WEIZMANN.BITNET@WISCVM.WISC.EDU
Message-ID: <860122164158.237078@DOCKMASTER.ARPA>
The excerpts from the ET-FIG news Vol 1 Number 5 contained an entry that
calculates the desired number of digits of Pi. However, upon
examination of the provided code, it only gives the digits of the
division:
355/113
While this is an approximation for Pi, it is not correct and in fact
differs from Pi in the 7th place past the decimal point. Pi is defined
as an infinite series and no simple division can provide accuracy to an
arbitrary number of places.
-- Doug
Acknowledge-To: <UMFORTH@WEIZMANN>UMFORTH@WEIZMANN.BITNET ("F.I.G.I.L.") (10/17/86)
========================================================================= Date: Tue, 7 Oct 86 14:44 EDT From: "Maj. Doug Hardie" <Hardie@DOCKMASTER.ARPA> Subject: Double precision floating point I have a LSI-11/2 with single precision floating point hardware. However, I need double precision (either 64 or 80 bits). Does anyone have the algorithms for F+, F-, F*, and F/ that could be used? It would be nice if they used the existing hardware instructions, but that is not necessary. -- Doug
UMFORTH@WEIZMANN.BITNET ("F.I.G.I.L.") (10/17/86)
From : JESUP RANDELL <JESUP@ge-crd.arpa> Date: 2 Oct 86 12:48 EST I am especially interested in 32-bit forth standards, and ways to work within a multitasking system cleanly (esp. without virtual memory). I have written a 32-bit, subroutine threaded forth for the amiga (68000 at 8MHz), that runs 10 iterations of the sieve in 9 sec, without using assembler. jesup@ge-crd.arpa