Out of curiosity I downloaded the VHDL source for the TG68K. My first reaction was, "Is that it? It's so small!"
When you consider the size of your typical software emulation implementation, it looks dinky 
I think the fact that the languages have defined microblocks ( I think I'm using the term properly) for basic operations (arithmetic, logic, data moving), these things have been preoptimised for performance. All one really needs to do in the TG68 core is decode the instruction, set the flags and pin signals correctly etc... Not to say that this isn't a huge achievement, just that the boring shuffling of bits is handled at a lower level, which saves a lot of space.
The whole tone of this post sounds negative and wrong... Try and reread it with a smile, as it's supposed to sound positive and excited...