Tap water probably would not work at all. Too conductive. EDM requires a potential to be built up between the workpeice and material. That potential is released in the form of a spark, the spark at some estimated 20,000+ degrees. That's why the fluid (water for wire, oil for sinker) is called a dielectric; it must have a calculated resistance to the flow of electricity. If it's too conductive, it will just short out and give you contact errors. If it's too resistive, it will not be able to jump a spark, or the machine will just run the electrode into the part and give you contact problems again.
As far as using water as a dielectric fluid for sinker, good luck. It's used with great success on small hole burning, like an EDM drill, but typicly does not work out well for large surface area work like a sinker. It simply cannot handle the heat and just vaporizes.
For the feedrates, there is so much more to EDM than power input. On time, off time, stabilizers, finish requirements, electrode wear, jump speed, flushing, flushing, flushing, and did I mention flushing? CNC sinkers will burn faster as they can use a larger spark gap for roughing and orbit the finish passes to get close enough for low IP sparks to clean up. This can become a real problem if you're doing 3d electrodes on a CNC mill, since you have to write two different programs, and less of a problem for simple square or round electrodes that can easily be sized. ZNC or other manual table machines have to run a much smaller spark gap, and run slower. The size and shape of your burn will also have a lot to do with it's speed. A tapered rib, for example, actually increases in surface area during a burn since it's constantly burning the sides and the tip, not just the tip. It will tend to slow down as it gets in a part.