RE: elapsed time

Yivi,
        This code works just fine.  I used the sleep() function to make
it wait and swapped the parms in %DIFF().

     Hbnddir('QC2LE') dftactgrp(*no) actgrp(*new)          
     D sleep           PR            10U 0 ExtProc('sleep')
     D   seconds                     10U 0 value           
                                                           
                                                           
     DBENCH            S               T                   
     DBENCHSECS        S              4  0                 
     DX                S              9  0                 
      /FREE                                                
       BENCH = %TIME;                               
       // a whole lot of program                           
       // many reads, calculations and writes              
       sleep(5);                                           
       BENCHSECS = %DIFF(%TIME:BENCH :*S);                 
       dsply bench;                                        
       dsply benchsecs;                                    
       *INLR = *ON;                                        
      /END-FREE                                            

Thank you,
Matt Tyler
WinCo Foods, LLC
mattt@xxxxxxxxxxxxxx


-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Yivi
Sent: Wednesday, November 16, 2005 8:17 AM
To: RPG programming on the AS400 / iSeries
Subject: Re: elapsed time

>
> Delete/comment the line "DBENCHSECS S 4 0" and compile, no errors?
>
No, no errors. But no change in the final result either.
Usually I define those external variables for clarity sake.
Is that a problem?

> Define BENCH as type Z (timestamp), not type T (time).
>
 Same result. I didn't think that should matter. Both data types have
seconds, and are valid for the %TIME and %DIFF functions, right?

> Also, you might want to increase DBENCHSECS to 5,0, because at 4,0

 I realize that at 4.0 I am constrained in how many seconds I can count
at,
but the programs I am testing take between 20' and 70' on each
iteration.
Since the benchmarking code won't go into production, I can do with 4.0
(or
5.0, or 10.0 for the matter; it's really not important).
 Regards,
 I.-