Distribution of time interval between successive interrupt requests

In majority of computer systems, the most time-critical tasks are performed by interrupt service routines. In some cases pooling methods are utilized, especially when I/O hardware is not capable of stand-alone operation. When these two approaches are mixed together conflicts are likely. Pooling procedure may not tolerate delays caused by interrupts and interrupt-driven procedures may not be able to wait until pooling driver finishes its job. In various systems some of the time-critical operations can be repeated or skipped, if CPU fails to service them in time. Statistical methods can be used to verify if system performance satisfies requirements. To apply them the distribution function of time interval between two consecutive interrupt requests is calculated on the basis of a simple theoretical model. The model is then verified by empirical measurements.