budd@arizona.UUCP (tim budd) (04/01/85)
Yes, I've maintained for quite some time that from the programmers viewpoint the best use of a parallel machine would be from a language, such as APL, where the programmer doesn't have to think about parallelism. I have even come up with a framework and algorithms for a compiler for such a machine (see Toplas, vol 6(3), July 1984). Unfortunately, nobody with the right sort of money (I refuse to work for DoD related research, and in the field of supercomputers that does not leave many others) has come forward to help in continuing this work. (I tried to get CWI, where I will be visiting next year and who have access to a 205, to fund this, but they seemed more interested in other things). So for the time being my ``APL Compiler for a Vector Processor'' remains a paper design.