now reading:
Retrospective on High-Level Language Computer Architecture [Ditzel and Patterson 1980]: a summary of failed design approaches for
- reduction of the semantic gap between programming and machine languages
- reduction of software development costs
- aesthetics ("esoteric")
[High-level language computers] are aesthetically appealing to those not familiar with modern compiler writing technology. It is acknowledged that code generation may be simpler for a high-level language computer. What needs to be made more fully understood is that a high-level language instruction set does not eliminate the need for compilers, nor does it greatly simplify them. The need and complexity of compilers extends far beyond code generation. The amount of code necessary for preprocessing, lexical analysis, syntax analysis, assembly, optimization, loading, error detection, error recovery and diagnostics often dwarfs the part of the compiler concerned with code generation. The level of the target computer does not seem to have enough of an effect on the size of a compiler to warrant a totally new architecture.
ref:
https://dl.acm.org/doi/pdf/10.1145/800053.801914#compilers #computerarchitecture #forth #retrocomputing
