Thanks JV,
Unfortunately, we want to read the pulswidth of the first pulse to come (hardware triggered), then do some things with it, do some other stuff in parallel and then read some more. So buffered reading isn't going to help much.
We started from the 'Measure Pulse (DAQ-STC).vi' sample, changing it to do triggered measurements and tweaking timeout and errorhandling to our needs.
It does seem that the 6020E has been a bad choice for the our task at hand (many single reads), as we've had performance issues in analog readings as well. Probably have to dump it all and go for some PCI solution.
Peter