Malcolm has started running into problems where get_point is unable to keep up with the pace of scans (now into the kHz range) - buffering can only help so much.
get_point has quite a lot of indexing logic to retrieve the values from the component generators based on tiled masks. Can anything be done to simplify this, perhaps by pushing the memory-speed tradeoff made previously further in the "more memory for more speed" direction?
Malcolm has started running into problems where get_point is unable to keep up with the pace of scans (now into the kHz range) - buffering can only help so much.
get_pointhas quite a lot of indexing logic to retrieve the values from the component generators based on tiled masks. Can anything be done to simplify this, perhaps by pushing the memory-speed tradeoff made previously further in the "more memory for more speed" direction?