samsonlonghair wrote:The problem is that FPGAs are expensive.
I guess "expensive" is kind of relative. From my perspective, FPGAs tend to be cheap for smaller batch implementations of custom ICs, and sometimes even high-volume. The alternative is usually an ASIC, which requires a lot more in development (actually laying out the transistors), and production resources; Only really becomes effective in high volumes, since the first chip is so exorbitantly expensive to manufacture. The FPGA is really most valuable when reconfiguring designs, since you can't really change the dies on an ASIC once they've been cast.
It's conceivable that someone could actually make a custom IC that replicated the N64, if he wanted to. The N64 used 350 nanometer transistor technology, and a lot of processes now are 10-nm, which is sort of nebulous in that there are a lot of factors that go into transistor densities in these processes, and that's partly controlled by the layout. Anyway, even nearly ten year old, 35-nm processes should be more than enough to replicate the N64 on a single chip. Xilinx uses 45 to 16-nm technology, although that's not quite analogous, and won't perform as well as similar ASICs, in terms of "density" and speed, due to the overhead from the functions that make the transistor interconnects reconfigurable, but distributors sell FPGAs with microprocessers preloaded into a fraction of the chip. I can hardly conceive of an FPGA these days that couldn't handle the
almighty power of the N64.
I realize this isn't entirely germane to the topic, and not everyone necessarily cares all that much about these kinds of technical details, but I guess FPGAs are getting more common in these niche consumer products. They can be kind of mystifying devices for anyone who hasn't worked with them before, but they're not quite as amazing as they might initially seem. Although, yeah, a lot of the typical FPGAs out there will run you $100 or well more, per chip.