Problems with large 4GL reports

I have been having problems running 4GL reports on large data sets. If I run the same reports on smaller data sets they work fine, but when I push all my data through, the report runs for about an hour before dying with an SQL error -408. I have had previous problems scaling my 4GL reports, and those either had to do with too many USING statements or too much temp table usage. In this report, however, I thought I was careful to avoid both problems. This report was working up until I added a rather large SELECT SUM(...) statement into one of ORDER BY levels of the report, but I have another report that conains more USING's, just as much temp table usage, and the same large SELECT statement, only the other one works. This has left be pretty dumbfounded. I am not even sure what issue is preventing me from scaling up my reports, so right now I am shooting in the dark to try and debug this. As an interesting sidelight, when I reduce the amount of computation in the report, the report will run longer before finally dying. It came pretty close the last time I went through and stripped out all the nonessential code I could find, but it's about as lean as I can make it and it didn't quite get all the way through. Is there some kind of fixed buffer to hold 4GL calculations that I am overrunning? I'm pretty dry on ideas of what might be going on.