IC5Notice: We have upgraded developerWorks Community to the latest version of IBM Connections. For more information, read our upgrade FAQ.
Topic
  • 7 replies
  • Latest Post - ‏2012-05-01T15:19:31Z by pelderon
PavankumarM
PavankumarM
10 Posts

Pinned topic Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

‏2012-04-14T17:54:10Z |
 One of our Customers is running Business Logic in Enterprise PL1 for zOS  , IBM PL1 for AIX , Open PL1 on Windows . Customer is looking at consolidating all Business Logic in IBM PL1 for AIX . Towards this wanted to understand a) Differences in   FLoating Point operations between the three PL1 implementation b) Differences in Levels of Precision provided by the three PL1 Implemntation .   In General any documentation on comparisons between the implementations (atleast  zOS , AIX) would be usefull in highlighting the risks to the customer
Updated on 2012-05-01T15:19:31Z at 2012-05-01T15:19:31Z by pelderon
  • pelderon
    pelderon
    11 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-04-16T20:26:16Z  
    PL/I on Z/OS supports hexadecimal float, IEEE float binary, and IEEE float decimal. However, PL/I for MVS and earlier PL/I compilers supported only hexadecimal float, and hence almost all float data used by z/OS PL/I programs is in hexadecimal format. PL/I on AIX supports IEEE float binary. These 2 formats all have a short, long and extended format, but the numbers have different values for the exponent ranges supported and for the values held in the mantissa. There are more details in the Language Reference Manual.
  • PavankumarM
    PavankumarM
    10 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-04-19T11:41:51Z  
    • pelderon
    • ‏2012-04-16T20:26:16Z
    PL/I on Z/OS supports hexadecimal float, IEEE float binary, and IEEE float decimal. However, PL/I for MVS and earlier PL/I compilers supported only hexadecimal float, and hence almost all float data used by z/OS PL/I programs is in hexadecimal format. PL/I on AIX supports IEEE float binary. These 2 formats all have a short, long and extended format, but the numbers have different values for the exponent ranges supported and for the values held in the mantissa. There are more details in the Language Reference Manual.
     Are PL1 Macros supported in IBM PL1 for AIX
  • pelderon
    pelderon
    11 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-04-19T15:44:37Z  
     Are PL1 Macros supported in IBM PL1 for AIX
     yes, the macro facility is identical across PL/I for z/OS. AIX, and Windows
  • PavankumarM
    PavankumarM
    10 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-04-29T13:59:21Z  
    • pelderon
    • ‏2012-04-19T15:44:37Z
     yes, the macro facility is identical across PL/I for z/OS. AIX, and Windows
    Can  IBM PL1 on  AIX emulate HEX FLoat on zOS PL/I.  . Does IBM PL1 on AIX have the same Run Time Library as IBM COBOL on AIX
  • pelderon
    pelderon
    11 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-04-30T21:12:14Z  
    Can  IBM PL1 on  AIX emulate HEX FLoat on zOS PL/I.  . Does IBM PL1 on AIX have the same Run Time Library as IBM COBOL on AIX
     Yes, IBM PL/I can emulate hex float on AIX, but this will not perform as well as using native float.
     
    And PL/I for AIX has a separate library from COBOL for AIX
  • PavankumarM
    PavankumarM
    10 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-05-01T03:26:47Z  
    • pelderon
    • ‏2012-04-30T21:12:14Z
     Yes, IBM PL/I can emulate hex float on AIX, but this will not perform as well as using native float.
     
    And PL/I for AIX has a separate library from COBOL for AIX
     Thanks for the clarifications . Currently the Customer uses CMPAT(V1) as the default COmpiler option while compiling Programs on zOS .  Does this indirectly corespond to HEX Float usage
     
    For High precision requirements , precompiler options(LIMITS( FIXEDBIN(31,63),FIXEDDEC(15,31))) is used . Wanted to understand how this improves precission as I understand that Floating Point Binary/ Decimal is not in effect
  • pelderon
    pelderon
    11 Posts

    Re: Difference in FLoating point operations between Enterprise PL1 , IBM PL1 for AIX , Open PLI

    ‏2012-05-01T15:19:31Z  
     Thanks for the clarifications . Currently the Customer uses CMPAT(V1) as the default COmpiler option while compiling Programs on zOS .  Does this indirectly corespond to HEX Float usage
     
    For High precision requirements , precompiler options(LIMITS( FIXEDBIN(31,63),FIXEDDEC(15,31))) is used . Wanted to understand how this improves precission as I understand that Floating Point Binary/ Decimal is not in effect
     the CMPAT option has nothing to do with float usage in any way
     
    the FIXED BIN and FIXED DEC limits also have nothing to do with float usage. However, they clearly allow for more digits in FIXED computations