I can take a stab at that. The perf is fine for a lot of things. The code on the other hand is kind of 'odd' to follow. It was a C project. #define everywhe

I can take a stab at that. The perf is fine for a lot of things. The code on the other hand is kind of 'odd' to follow. It was a C project. #define everywhere to basically create an object like system. They now are moving to C++ in some places they use it to good effect and stripping out that #define system. Other spots are still the same C code from 15 years ago. So now you have a project that is both. So it makes it sort of strange in a few places. They have over the years made decent progress to strip out some of the foot guns that C lets you do. Then on top of that 6 different project leads with differing opinions on how to run the project over the years. They have also smashed together two different 'ROM' systems. One is a static system where everything is burned into the code and the other is a dynamic system using XML. So context burden is higher when trying to change anything. Add to that unless you know their system building a new emu is kind of interesting. But it is still sort of funny seeing people rocking a .34 version. 'because of performance'. When anything within the last year is probably just as performant and wildly more capable and accurate. If the chart I saw last week is any indicator 241 should have at least a 2-5% speed bump because they removed a C'ism and flipped it over to use a C++ style parameter in some core calls. Most of this goes back to around 110 or so when they flipped over to use C++. Well they also added in a decent regression in speed at that time too. So those two things became conflated in peoples mind that C++ bad. When there was a design issue. Good place to follow the code and see what is going on. https://github.com/mamedev/mame https://github.com/mamedev/mame I would estimate they bash out about a dozen pull requests a day.