SUPer has always kept an internal copy of the events graphics in RAM. It was just easier to debug and process that way. However this causes a dramatic memory usage on long epochs with many events: it is not unusual for SUPer to allocate up to 2 GB of dynamic memory.
While the process of caching the graphics does speed up the conversion process in general, it prevents the usage of SUPer on extraordinarily long epochs with tens of thousands of events. This PR gets rid of the caching structures altogether.
SUPer has always kept an internal copy of the events graphics in RAM. It was just easier to debug and process that way. However this causes a dramatic memory usage on long epochs with many events: it is not unusual for SUPer to allocate up to 2 GB of dynamic memory.
While the process of caching the graphics does speed up the conversion process in general, it prevents the usage of SUPer on extraordinarily long epochs with tens of thousands of events. This PR gets rid of the caching structures altogether.