Re: Decimal versus binary arithmetic was Re: J4 - presentation/discussion on "Fu

On Wed, 2 Apr 2008 18:51:00 +0000 (UTC), docdwarf@panix.com () wrote:

>In article <WqCIj.292096$Gl5.205581@fe02.news.easynews.com>,
>William M. Klein <wmklein@nospam.netcom.com> wrote: 
>
>Given:
>
> WORKING-STORAGE SECTION.
>  01  NUMERATOR VALUE 1 COMP-5 PIC 9(9).
>  01  DENOMINATOR VALUE 5 COMP-5 PIC 9(9).
>  01  QUOTIENT COMP-5 PIC 9(9)V9(4).
>*
> PROCEDURE DIVISION.
>*
>     COMPUTE QUOTIENT = NUMERATOR / DENOMINATOR.
>     DISPLAY QUOTIENT.
>     GOBACK.
>
>... compiled with
>
>//COB      EXEC PGM=IGYCRCTL,
>//            PARM='SIZE(MAX),LIB,MAP,DYNAM,LIST,
>//             XREF,OPT,FLAG(I,I),NUMPROC(PFD),'
>
>... and using PP 5655-G53 IBM Enterprise COBOL for z/OS  3.4.1, the result
>I get is:
>
>000012  COMPUTE
>   0002B6  5840 912C               L     4,300(0,9)
>   0002BA  5830 4008               L     3,8(0,4)
>   0002BE  1F22                    SLR   2,2
>   0002C0  5870 4000               L     7,0(0,4)
>   0002C4  1F66                    SLR   6,6
>   0002C6  5D60 C000               D     6,0(0,12)
>   0002CA  4E70 D108               CVD   7,264(0,13)
>   0002CE  F154 D0FC D10B          MVO   252(6,13),267(5,13)
>   0002D4  4E60 D108               CVD   6,264(0,13)
>   0002D8  9110 D101               TM    257(13),X'10'
>   0002DC  D204 D101 D10B          MVC   257(5,13),267(13)
>   0002E2  4780 B0BE               BC    8,190(0,11)
>   0002E6  9601 D105               OI    261(13),X'01'
>   0002EA                 GN=15    EQU   *
>   0002EA  D203 D0F8 C008          MVC   248(4,13),8(12)
>   0002F0  F070 D0FE 0004          SRP   254(8,13),4(0),0
>   0002F6  5D20 C000               D     2,0(0,12)
>   0002FA  4E30 D118               CVD   3,280(0,13)
>   0002FE  F154 D108 D11B          MVO   264(6,13),283(5,13)
>   000304  4E20 D118               CVD   2,280(0,13)
>   000308  9110 D10D               TM    269(13),X'10'
>   00030C  D204 D10D D11B          MVC   269(5,13),283(13)
>   000312  4780 B0EE               BC    8,238(0,11)
>   000316  9601 D111               OI    273(13),X'01'
>   00031A                 GN=16    EQU   *
>   00031A  FDD5 D0F8 D10C          DP    248(14,13),268(6,13)
>   000320  F897 D120 D0F8          ZAP   288(10,13),248(8,13)
>   000326  960F D129               OI    297(13),X'0F'
>   00032A  D207 D118 A074          MVC   280(8,13),116(10)
>   000330  D204 D11B D125          MVC   283(5,13),293(13)
>   000336  4F20 D118               CVB   2,280(0,13)
>   00033A  F144 D11B D120          MVO   283(5,13),288(5,13)
>   000340  4F70 D118               CVB   7,280(0,13)
>   000344  5C60 C000               M     6,0(0,12)
>   000348  1E72                    ALR   7,2
>   00034A  47C0 B126               BC    12,294(0,11)
>   00034E  5A60 C004               A     6,4(0,12)
>   000352                 GN=17    EQU   *
>   000352  1222                    LTR   2,2
>   000354  47B0 B130               BC    11,304(0,11)
>   000358  5B60 C004               S     6,4(0,12)
>   00035C                 GN=18    EQU   *
>   00035C  9067 4010               STM   6,7,16(4)
>000013  DISPLAY

Obviously on an IBM z series system the last thing you want to do is
scaled arithmetic in binary.  This makes me even more curious to see
what gyrations are done by the p series compilers and any other
compilers which will show generated instructions for basically binary
only machines.  That code is SLOW.
>
>... so it would seem to ConVert to Decimal (0002CA, 0002D4, 0002FA,
>000304), Divide Packed (00031A), Zero and Add Packed (000320) and then
>Convert to Binary (000336, 000340)... along with a bunch of Other Stuff.
>
>(What comes next, of course, is Mr Klein's response along the lines of
>'How interesting... could you compile it again with options SNKRS, LNCHBAG
>and NWSPPR and post what that does?')
>
>DD

Clark Morris