It is a TON of work to do this. Not easy by any means. I doubt this will become anything more than a cutesy one or two a year, at most.
I thought it was cool.
The real advantage would be that you could emulate the hardware more accurately including detailed timings.
One of the achilles heels of emulation is that you're doing other stuff on the CPU and slowdown and other problems (such as cache performance) can compromise the timing of the emulation.
I'm sure this could be mitigated with better code, but I can't help but think that FPGA implementations have the ability to more faithfully represent the hardware than a CPU-based emulation. maybe when FPGA's come standard with all computers...
--
- Ahigh
This is very cool yep, and I agree doing it in HW probably allows for more accuracy. The speed of computers seems to be getting to the point where going HW doesn't have as much advantage as it may have in the days of Sparcade (486 or so). One of the problems w/ FPGA impls seems to be completion, much like a lot of emulation projects. The C-One is a good example. From what little I know about it the goal of having a complete HW simulation of a Commodore 64 is slow going. Not sure how well full computers compare to arcade pcbs but I bet there are a lot of chips to emulate in either case.
Firmware upgrades for the FPGA would be the new equiv of MAME upgrades.

The cool thing lately has been how this type of work has allowed for stuff like the C64 DTV. Very small portable but ACCURATE game machines. I also heard on retrogamingradio in an interview w/ Jeri (the creator of the C64-DTV) that shw is working on an FPGA that emulates a lot of the old arcade games and will be part of a mame cabinet type of deal that will sell in places like Wal-Mart. If it turns out anything like the C64-DTV it will be very accurate.
~telengard