Our customers have given us great feedback on the high performance of our JamaicaVM runtime and AOT compiled code. However, only a few customers have provided us with formal or informal measurement results. A couple of these results are available below. If you have run your own benchmarks with JamaicaVM, with the benchmark code provided below, or other benchmarks,and are able to provide the results publicly, we would like to publish these here. Please let us know if you are interested in doing so.
The aicas Jitter test is a set of tests that runs periodic tasks using different means provided by Java and Realtime Java.
The test perfoms a small task involving lists and memory allocation on every release. The test is run for a given number of periods. A run is successful if no overrun occurs and if the release time Jitter is less than 50%.
After a successful run, the period is reduced by a factor and the number of iterations is increased by the inverse factor and the test is repeated for this new period. This continues until a period is found that was not successful.
This is a small benchmark to measure the scheduling latency between two realtime threads of control using two RTSJ AsyncEventHandlers. The first handler is attached to a periodic timer. It fires a second event thereby releasing the second event handler. The time between releasing the event attached to the timer and releasing the second handler is measured to determine the time it takes to transfer control from one thread context to another, hence the scheduling latency.