It would be good to have some performance benchmarking tests, to ensure that performance is never degraded with new releases.
For example, adding the ability to trap CPU halt is not likely to reduce performance, as the additional work is being done at a time when, by definition, the emulator is entering a low-workload state. On the other hand, trapping each CPU interrupt (eg. for acceptance tests to hook into) could adversely affect performance as many interrupts are triggered every second.
It would be good to have some performance benchmarking tests, to ensure that performance is never degraded with new releases.
For example, adding the ability to trap CPU halt is not likely to reduce performance, as the additional work is being done at a time when, by definition, the emulator is entering a low-workload state. On the other hand, trapping each CPU interrupt (eg. for acceptance tests to hook into) could adversely affect performance as many interrupts are triggered every second.