Am in full agreement with jacquesm.
Here's my partial list:
Application virtualization, <2000. Submitted invention disclosure in 2003 or so, when prior art search revealed someone had a patent issued a year earlier (so they must have been filed several years before that).
Physics hardware acceleration, 2000, while working on a virtual reality game. Learned about PhysX in about 2006 or so.
RISC and VLIW processor architectures, 1996, junior year undergrad, while not being aware of their prior existence.
Following two were while I was working on Sinclair Spectrum+ at 12-15 years of age, and largely unaware of what already existed at that time:
3D computer graphics/animation, 1990, 13 years of age, after learning coordinate geometry two years ahead of peer students. Tried to come up with a simple mathematical function f(x, y) that that would resemble a human face, with no success. Prototype created frames offline using interpreted BASIC and cycled them on the screen using hand-written machine code.
Compilers, while using slow interpreted BASIC on Sinclair Spectrum, 1992, 15 years of age, when I first learned how to program in machine code manually. My prototype would only convert text code in reverse-polish notation (RPN) to RPN instructions available on the processor (so more like assembler really).
Envisaged zero static power logic with complementary relays (like CMOS), at high school, while not knowing anything about CMOS then.
Lesson learnt: (See jacquesm's comment).