When it comes to subgenres cyberpunk has to be one of the most varied. It's origins are rooted conceptually in the novel "Neuromancer" by William Gibson, and visually in the film "Bladerunner." Oddly enough, most noteworthy entries don't really follow the format very closely (if at all). "Akira," for example, has the moodily lit city at night, but forgoes bionics and cyberspace for bikers and psionics. Another film, "Inception," has technology that allows people to share dream-spaces (basically high-res versions of cyberspace), yet lacks any of the signature visuals typically associated with cyberpunk. Cases such as these make it hard to really define the conventions of the subgenre. The best summary I've ever heard is, "High tech meets low life." Almost every cyberpunk story involves criminal activity in a big way. The technology on display also tends to be the kind of thing that might actually be possible sometime in the foreseeable future. Granted, what actually becomes viable down the road is far from certain. Cybernetic enhancements, wherein mechanical limbs are superior to organic ones, are still a heck of a long way off. "Ghost in the Shell" has the concept of a human brain in a robot body. At first it might sound like a plausible scenario not-so-many years from now, but it quickly becomes silly once you consider all that grey matter needing oxygen supplied by red blood cells, which in turn must be replenished by bone. On top of that nutrients must be supplied which means stomach, intestine, liver and kidneys. Replacing certain internal organs, such as the lungs and heart, with artificial substitutes is currently within the realm of possibility, but there's no way anything (short of a sprawling chemical refinery) can do the job of a human liver. It's a common failing of science fiction writers to drastically undervalue the complexity or, for lack of a better term, "engineering precociousness" of the human body.
People uploading their consciousnesses into a computer is also one of those cases of "it might as well be magic." Our current understanding of the human brain is pretty limited. Even coming up with a way to get an accurate picture of the wiring, let alone copying it, is something that still eludes neuroscientists. I have a feeling that even if they were to crack that particular nut, it would be incredibly difficult to translate all those neural pathways and connections into the binary language of computers. At the very least I would be incredibly surprised if the file size for a person's brain came out to anything less than hundreds (if not thousands) of zettabytes of data.