Custom simulation is really slow already at the moment. To make sure it does not get even slower, create some benchmarks and write tests for them to make sure that a random change does not break down the overall performance massively.
We do not even need to deal with nasty and device-dependent time values here: Just run the relevant code in something like a Sandbox and count the number of primitives executed. This would be stuff enough for a separate tool. :D
Custom simulation is really slow already at the moment. To make sure it does not get even slower, create some benchmarks and write tests for them to make sure that a random change does not break down the overall performance massively.
We do not even need to deal with nasty and device-dependent time values here: Just run the relevant code in something like a Sandbox and count the number of primitives executed. This would be stuff enough for a separate tool. :D