This problem is somewhat more tolerable (eg. you CAN make it work if you are patient enough to keep repeating the test until the CPU Emulator DOESN'T crash) if you:
1. Run at a slow speed.
2. Avoid switching tasks while it's running.
You still get big huge long Java stack traces that talk about NPE's and StrategyPaintManager and so on, but, it seems to work anyway. But, between #1 and the necessary retrying, it can take a half-hour, and oftentimes you find a 'real' problem and have to repeat the whole process anyway.
Obviously these symptoms sound like threading problems. It's ironic, or maybe not, that the book doesn't get into threading and interrupts. Oh, and lastly, I run Java apps all the time, some of them monsters like Eclipse, and none of them have problems remotely as bad as your simulator.