In a message dated 7/19/2005 9:22:04 A.M. Central Daylight Time, [EMAIL PROTECTED] writes:
Yet, I prefer "L Rx,=A(equated-symbol)" so the equated symbol may be used in other contexts, such as storage declarations. Another reason why I also prefer this technique is so the equated symbol can be found in the cross reference dictionary or by doing a FIND command. E.g., if you code LA Rx,ONE or L Rx,=A(ONE) and you search the code for all occurrences of the character string "ONE", you will probably find a lot fewer than if you search the code for all occurrences of "1", almost all of which will not be relevant (e.g., references to R1, R10, R11, ..., R15). If later you need to change the value being loaded into Rx from 1 to 2 you can very quickly find LA Rx,ONE and change it to LA Rx,TWO and then reassemble (after adding an EQU statement for TWO, of course). If having the literal operand in a different page which may cause a page fault at execution time is a significant performance problem, then the literal can be forced to be close to the instruction referencing it through several techniques. I don't see an extra page fault as a big performance hit unless the code is being executed a huge number of times per second. It is a fine point of programming elegance to avoid page faults. It is usually much more important to get the module debugged and out the door as fast as possible, since programmer costs per hour are a lot more than the cost of one page fault. Bill Fairchild ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
