When I measure the interval from the start of trial t1 until the start of the next trial, I get 250msec.
How exactly did you measure that?
Ist such a preparation time to be expected on a 2,4 GHz Intel Pentium 4 with Inquisit running at realtime priority?
Yes, that can happen. Inquisit is designed to nail timing *within* a <trial>, not necessarily *between* trials. Here:
(a) Time is needed to prepare stims before the stimulus presentation sequence and response polling can start. There is no /pretrialpause [1], so that happens sometime between the trial is first initialized. Once the stims have been readied (memory allocated, etc.) stimulus display will start with the next available frame. So, there goes (at least) one frame into your ITI.
(b) Similarly, some time is needed for some clean-up (de-allocate memory, clear response buffers, etc.) at the end of the trial / before the next trial can start. There goes another frame (under optimal conditions). As above, to make this interval more predictable / control it, one should use /posttrialpause if possible.
(c) Other stuff may be going on such as erasing (i.e. overwriting) certain visual stims, but that's not possible to say with certainty based on your code snippet.
[1] Quoting from the "How to Control Trial Duration and Inter-Trial Intervals" topic re. pretrialpause: "Pauses for the specified duration at the begining of a trial, prior to stimulus
presentation. In addition to providing a general means of controlling
inter-trial intervals, the PretrialPause is useful for experiments that present
large numbers of memory intensive stimuli on a given trial. Depending on the
size of the stimuli and the speed of the hardware, stimulus preparation may add
notable lengths of time to the beginning of the trial. Furthermore, stimulus
preparation time may vary from trial to trial, in which case varying durations
may be added to the beginning of the trials. However, if a PretrialPause
interval is specified, Inquisit uses this time to prepare the stimulus
presentation sequence. By specifying a PretrialPause duration long enough for
stimulus preparation to complete, the experimenter can impose a constant and
predictable duration at the beginning of each trial."