Im surprised at the number of replies to this thread... so ill add a couple more thoughts.
1. I wonder how successful ASIC Scrypt miners will really be.
2. Im currently working on my own FPGA implementation, learning the backend algorithm is kinda fun.
I would imagine they should be about the same once they're programmed to work on scrypt algorithms.
Some say SHA-256 is simpler then Scrypt...so maybe the ASIC chips will be slower? IDk lol.
I don't believe that is the case. I have not spent any time trying to optimize SHA-256, but from what I understand it is an algorithm that is naturally parallelized and as such custom ASIC's can be produced to crunch these numbers faster. Scrypt on the other-hand was specifically designed to make it costly to perform large scale custom hardware attacks by requiring large amounts of memory. This doesn't mean it is impossible to come out with better custom Scrypt hardware, however those that are successful at it will have to be more creative than SHA-256 designers. I'm currently working on it, however I'm more doing it for fun than for profit. I have a day job, so my time spent on this is fairly limited.