David Crayford wrote: >REXX performance is terrible. OK, now we're starting to get somewhere!
>It's used extensively on z/OS, a platform where CPU >consumption costs big bucks. 1. Start with REXX compilation. 2. How much is "extensively"? What is your resource consumption attributable to compiled REXX at monthly peak? Let's suppose that number, reasonably carefully determined, is "X" MSUs, "P" percent of your total peak MSUs. 3. How do "X" (and "P") change, including zIIP effects, if the compiled REXX code that is capable/compatible runs instead via NetRexx with the latest JVM? 4. For the remaining compiled REXX code that is not capable of running via NetRexx, how much could "X" (and "P") be feasibly reduced if there were an improved compiler (JIT or otherwise)? Let's quantify this. -------------------------------------------------------------------------------------------------------- Timothy Sipples IT Architect Executive, Industry Solutions, IBM z Systems, AP/GCG/MEA E-Mail: [email protected] ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
