The Afterburn module can add some really nice low end character and scooped focus on frequencies; however, the cpu damage is fairly high. Could it be optimized? Could certain parts be ignored – if it was reconsidered in terms of the “problem” it addresses, are there alternative approaches? Or just: Are there any parts that can be replaced; are there nodes strung together that could be replaced by a different set of nodes and math expressions?
I had a look at the internals of the Afterburn and most of the CPU load is caused by the filters included in the module. It uses a combination of a low shelf, peaking and high shelf biquad filter as a three band equalizer and has one for each channel both before and after the distortion module for a total of 12 biquad filters. Since the biquad is pretty efficient there’s no much room for improvement there. You could potentially remove the filters either before or after the distortion modules but I’m sure that would change the character of the output considerably. Another option would be to reconfigure the unit for mono operation. It’s essentially 2 mono units packaged together so converting it to mono would reduce the CPU load by almost 50%. There might be some benefits in reconfiguring the distortion module. It’s pretty expression node heavy and some could probably be consolidated, but any gains would be minor.
I’m not totally certain what the overall intent of this module is so I hesitate to suggest alternatives. It seems somewhat complex for a distortion unit but the inclusion of the equalizers gives it more control over the sound than a simple unit would provide. Separating the filtering and distortion into individual modules would give you more flexibility but you would lose the “all-in-one” convenience of a combined unit.