...running coldfire code. It won't be anywhere near that when running typical 68k code.
If I remember correctly, if at least the exec library was compiled for coldfire (which is now possible thanks to the kickstart bounties) emulation would be faster (and if the whole kickstart is recompiled it should be even faster).
But I agree that nowadays an ARM cpu (with built in opengl es 2) would be a far better choice, even if it does not emulates the 680x0 in hardware, it is cheaper and faster. IMO a winning combination would be a board w/ARM cpu plus a FPGA that fully emulates any 68k.