After all this time that I've dedidated to produce DuoS, I can assure that there isn't any portion of a DS emulator more CPU consuming than its GPU (Graphics Processing Unit).
I've tried to optimize it in a lot of ways, without relaying on the host's (PC's) multithreading and GPU accelleration as much as I could... in order not to compromise emulation quality and compatibility. To be honest, regarding this... I already ran out of ideas.
On the other hand, since last night, DuoS has now a fully enabled Dynamic Recompiler core, and do not relay on "secondaty calls" to the Interpreter core anymore. This is translated into a huge speed up, as well as a very noticeable reduction of CPU usage in the host computer. It will be the work of our noble testers to determine how does this affect DuoS's compatibility... but we will soon find out.
In the meanwhile... I see a couple of days of continuous debugging coming my way...