IBM Mainframe Forum Index
Log In
IBM Mainframe Forum Index Mainframe: Search IBM Mainframe Forum: FAQ Register

Speed-up/reduce CPU use of code

IBM Mainframe Forums -> PL/I & Assembler
Post new topic   Reply to topic
View previous topic :: View next topic  
Author Message
Warnings : 1

Senior Member

Joined: 07 Feb 2009
Posts: 1241
Location: Vilnius, Lithuania

PostPosted: Tue Sep 26, 2017 12:05 am
Reply with quote

One of the programs I'm responsible for requires some new functionality, Top-nn tables, at least one summary set of three (Top-50) of distance, time, and velocity, but potentially rather a lot more, like the same set (but only as a Top-10) for each year (currently 36), each type (currently 18), country (currently 29 + 1), and nationality (currently 89).

Due to the fact that the number of items to process is unknown, all data is held in linked lists, but the sort is done on an, obviously controlled, array of of pointers to the various items in said linked list. So far, so good, this works like a charm for the summary tables, just take the first 50 items, put them into a table, and "Bob's your uncle..."

Producing the other Top-nn's can use the same sorted arrays, and that's what I'm doing at the moment, and it works flawlessly, but it's not exactly efficient, as there are sets that contain just a few items, in which case, should you wonder, the Top-10 might be reduced to a Top-7, Top-3, or in rather more than a few (27!) cases, a Top-1...

And those are the ones that are highly inefficient, as I have to go through almost the entire array to find the first item using the following:

#h = hbound(distance.d_ptr, 1);
#i = #h;

do c = 'K', 'T', 'V';
  #t = 0;

  do #i = #h to lbound(distance.d_ptr, 1) by -1 until(#t = #top_nn);
    list_ptr = distance(#i).k_ptr;  /* or time.t_ptr/velocity.v_ptr */

    if sty_tcn = 'N' & tcn = this_list.nat  | /* 89     */
       sty_tcn = 'Y' & sty = this_list.year | /* 36     */
       sty_tcn = 'C' & tcn = this_list.cnty | /* 29 + 1 */
       sty_tcn = 'T' & tcn = this_list.type | /* 18     */
       sty_tcn = '*' then                     /* 1 (summary) */
        #t = #t + 1;

        if #t = 1 then
            if c = 'K' | sty_tcn = '*' then
                #h = #i;
                write file(outfile) from(pline);
                write file(outfile) from(top_nn_sep);
                write file(outfile) from(top_nn_head);
                write file(outfile) from(top_nn_sep);
        line           = top_nn_init;
        top_nn_line    = this_list, by name;
        top_nn_line.v  = round(this_list.vx, 1);

I do a little bit of caching (actually a hell of a lot for the what turn out to be the Top-1 tables) by saving #i in #h for #t = 1 & c = 'K', but can someone suggest an even more efficient way. And no I don't want to sort individual sub-lists, as that's more than likely going to cost way more CPU than the current approach.

Any ideas?

Back to top
View user's profile Send private message
View previous topic :: :: View next topic  
Post new topic   Reply to topic View Bookmarks
All times are GMT + 6 Hours
Forum Index -> PL/I & Assembler


Similar Topics
Topic Forum Replies
No new posts Execute step regardless of previous s... JCL & VSAM 2
No new posts Error 0C1 Reason Code 1 with branch i... PL/I & Assembler 3
No new posts JCL for replacing code in Cobol JCL & VSAM 7
No new posts REORG to reduce segment size new DBD IMS DB/DC 2
No new posts Easytrieve EZABX007 with error code 0... CA Products 7
Search our Forums:

Back to Top