>I know that important applications for parallelism exist—rendering graphics, breaking codes, scanning images, simulating physical and biological processes, etc. But all these applications require dedicated code and special-purpose techniques, which will need to be changed substantially every few years.
Interesting... not that it pays to go against The Knuth, but surely this is the same argument that's been made countless times before when it comes to hardware advances?
It seems to me that someone will come up with a compiler/code technique to present a uniform interface to these "special-purpose techniques", the way OpenGL did for graphics, or MapReduce for multiprocessing or TCP for network access, or.. and who was it who said that essentially every advance/new technique in CS comes from indirection?
'the idea of... "unit tests" appeals to me only rarely... lots of time is wasted on activities that I simply never need to perform or even think about. Nothing needs to be "mocked up."'
Interesting... not that it pays to go against The Knuth, but surely this is the same argument that's been made countless times before when it comes to hardware advances?
It seems to me that someone will come up with a compiler/code technique to present a uniform interface to these "special-purpose techniques", the way OpenGL did for graphics, or MapReduce for multiprocessing or TCP for network access, or.. and who was it who said that essentially every advance/new technique in CS comes from indirection?