This is more of an FYI than a request for help, but Iím interested to see if other folks have run into similar problems. We have a VB form for my AUT which has these characteristics :
6 variations of the main form stacked up with a splitter between the controls and a results grid Ö and a picture box that brings one instance to foreground at any one time.
Each variation has 10-30 different controls on it, so max there are ~ 100 controls on the form, but only 20- 30 visible at any one time This was done to actually increase refresh rate and performance on the form itself when presented to the user, and works well.
Unfortunately the size of the form and its related objects/ controls makes Robot manipulation of it fairly slow, takes ~> 1 second between each Robot command even when I have the Playback option delay between commands set to 0 ms. Ie Robot is hideously slow at extended, data driven tests.
The developer of the form and I experimented by building an instance of the form with just 30 visible controls, no hidden controls. Robot will manipulate this form as fast as desirable (i.e. so fast that itís a blur), so this points to the number of controls on the real form as the culprit.

Usually I run into the opposite problem, that I have to slow Robot down with delayFor's or checkpoints to avoid going faster than the application- under- test. The test solution we have reached for this app is to build some thin forms as in the experiment above, to exercise the database and middle layer, then do the GUI verification on the fat form. This will work well, and presently I am reluctant to open an issue with Rational that will result in a custom ďsolutionĒ, since that may take time away from my primary objective, which is to test the AUT.