Newsgroups: comp.sys.acorn
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!rusmv1!helpdesk.rus.uni-stuttgart.de!zrzm0111
From: zrzm0111@helpdesk.rus.uni-stuttgart.de (MUFTI)
Subject: integer division, operation systems
Message-ID: <1991Jun27.191439.28073@rusmv1.rus.uni-stuttgart.de>
Originator: zrzm0111@helpdesk.rus.uni-stuttgart.de
Sender: news@rusmv1.rus.uni-stuttgart.de (USENET News System)
Organization: User Helpdesk Comp. Cent. (RUS) U of Stuttgart, FRG
Date: Thu, 27 Jun 1991 19:14:39 GMT
Lines: 86

Subjects: Integer division, CLI-editors, !formula, other Operation system as
          RISCOS

Hi 

Integer division:

 I tested the 3 routines of signed 32-Bit/32-Bit-integer-division I know
 (unrolled versions of Conrad Hughes's and mine's) and counted the 
 required cycles.

 This are the conditions to compare the counting the cycles:
 The two operands are in the right registers.  
 No check for division by zero is counted.
 No save/load of required temporaery registers is performed.
 All other required operations are counted (also fixing signs).

 The results:

 routine for      required Cycles                Bytes   
 integer c=a/b    best_case       worst_case           
 ===================================================== 

 routine of       107             107            428
 Conrad Hughes 

 routine of       29              201            224
 Acorn ANSI-C

 routine of       170             170            680 
 MUFTI

 =====================================================
 in my opinion:
 requries Acorns ANSI-C-routine more cycles if absolute value of b is less
          =====================

                          more Cycles      \
                   ------------------------->   
                                           /

       abs(b)>&FFFFFF, abs(b)>&FFFF, abs(b)>&FF, abs(b)>0
 
 Conrad Hughes's and mine routine was posted to comp.sys.acorn in past.


CLI-editors:
 
 In a further posting, Paul Moore said, that a UNIX-compatible Desktop-editor
 is required.

 I think you can live with the CLI-editors (like VI- and EMACS-clones), they 
 are also callable from the desktop.
 (Either with set alias$@runtype_FFF or a programmable icon (like !prog)) 
 Some Editors require the opening of a textport ( with echo <28>.... ).

!formula:

 Has ever someone tested !formula ? ( no response yet ...)

other operation system as RISCOS:

 Cause currently RISCOS has a lack of modern features ( all other 
 operation systems (accept the old operationsystems CPM,MSDOS,GEM and 
 AppleIIdos) are better in this cases ) like timesheduled multitasking and
 we need a better Operationsystem ( or a rebuild of main parts of riscos to
 allow such things ).
 I think the very old UNIX version II (=MINIX) is such a operation system,
 but is there still someone, who work to implement it ?
 I heard nothing in last 2-3 month from Martin Patzel or from Albert Coelmans.
  

so long
MUFTI

 INTERNET:
 zrzm0111@helpdesk.rus.ini-stuttgart.de
 
 this is in JANET-notation:
 zrzm0111%de.uni-stuttgart.rus.helpdesk@earn-relay
 or  
 zrzm0111%de.uni-stuttgart.rus.helpdesk@nsfnet-relay
 
 
 BITNET:
 ZRZM at DS0RUS1I 
