Intel 8086

The most infamous proprietary VM in computer history, that throttled for so many years advances in computer technology, has been the i8086 instruction set, upon which the "PC-compatible" computers were built. Because this "industry standard" was so low-level, software could not be written to take advantage of other better faster simpler smaller cheaper hardware architectures.

You're not convinced that the i8086 instruction set is a virtual machine? See how the modern compatible CPU, like the AMD K6, actually emulates the 8086 instruction set on top of a RISC core.

To be fair, Intel shouldn't be the most blamed one for that: the misdesign of an operating systems that so deeply depends on the processor structure, the lack of free portable development tools, was not their fault (they did encourage it, though). The only thing Intel did was to continue produce ever faster hardware in response to a demand for a series of compatible processors. and benefit from the PC market having relatively low price due to standardness. IBM and Microsoft should be blamed, together with all the competing OS vendors, who all wrote proprietary hardware-dependent OSes, so that in any case, customers were taken hostages.

However stubborn the legacy design, free competition from free specifications allows Intel-compatible platforms to currently give the more bangs per bucks. Hence, even if hopefully and there's no reason why intel should not continue to produce quality hardware after the end of that sad i8086 compatibility episode. Pressure from competition will have them work better and lower their prices.

The morale of the story is that:

  1. A virtual machine designed for a portable interchange of programs should be high-level enough so that it could be implemented efficiently on just any present and future hardware.
  2. It should be free software, not proprietary software.