Patent-Troll Backstory - Higher Math Emulation

Migration of PROSE to mainstream PCs in the early 1980s was infeasible because of low speed and 16-bit memory segmentation. We therefore designed a "higher-mathematical computer architecture" at the Aerospace Corporation as a platform for semantic-intensive languages of this kind. This architecture, adapted into a workstation, was designed to host "language MetaComputers", which were hardware enhanced virtual machines emulated from microcode. MetaComputer development was automated using metacompiler technology based on the Tree Meta compiler-compiler. Such scaffolding was intended to automate the application-specific evolution of this high-level computer architecture.

This technology was spun-off into a microcomputer startup, International Meta Systems, intending to migrate the workstation design into a microcomputer chip empowering a PC-AT coprocessor product, the IMS Max2 Metasystem. The first IMS patent disclosure for this design was submitted in 1987. But fabrication problems in this first IMS chip resulted in a degraded clock speed of only half its design speed of 30 MHz.

Higher Emulation Validation

Nevertheless, even at this speed, Whetstone benchmarks of this emulated Fortran MetaComputer, showed it to be 10x faster than Intel 386 Fortran which was augmented with the 387 floating point chip, whereas the MetaComputer's FP was emulated from microcode. Thus this important test validated the HLL MetaComputer emulation theory. But the MetaCalculus goals of IMS were later abandoned. After a brief project to create a Smalltalk MetaComputer for the Max2, HLL emulation was abandoned altogether and IMS turned toward emulation of CISC microchips in the "Intel Knockoff" market, ending with bankruptcy and the famous "patent-troll" lawsuit with Intel.

This early attempt to change the hardware game had little chance to succeed, because Moore's law had a few more decades of circuit shrinkage before it ran into the "power wall" and the window of opportunity quickly vanished. But that is not the case now. There is no better alternative to the MetaComputer approach this time. Occam's razor will prove that out.