Advances in Computer Architecture

This book discusses some CPU/computer architecture designs and design principles which should be considered advanced not only for the year of publishing (1982) but for today and for the future also. The book starts with discussion of semantic gap between mechanisms presented by real programming languages and real computers. The book also introduces different measures which can be applied to binary programs e.g. program size, number of bytes transferred etc. It's easily seen that semantic gap impacts software reliability but what's more interesting advanced designs not only minimize this gap improving software reliability and reducing development effort but also improve all the measures of binary applications. The book proves this comparing real IBM System 360 Model 158 with imaginary PL/1 machine implementing subset of PL/1 language and more general SWARD machine. The numbers characterizing advances of the later two are very impressive. It seems that the book can be easily overwritten today with references to C or Java and Pentium or UltraSPARC without changing the conclusion. Some real designs are pictured: forgotten today iAPX432 (which seems had given i286 its memory management), research database machines etc. Author (mostly taken from RadiSys web site) - Glen Myers - President and CEO of RadiSys which he founded in 1987. Prior to that, he held management positions at Intel in the microprocessor components area. Among other things, he was responsible for the definition of the i386 and i960 processors, and he holds three patents on the chip design of the i960. Prior to joining Intel, Myers was with IBM for 13 years, at the IBM Poughkeepsie laboratory and the IBM Systems Research Institute. He is the author of eight books on computer architecture and software engineering (including often mentioned "The Art of Software Testing (1979)". The book is good written and easily read (at least in its Russian translation which is the only one I could read about ten years ago). Since I missed somewhere the first part of it (it was in two volumes in Russian) and don't have the second at hand probably I missed something and I'm not sure for correct spelling SWARD / SWORD for which I couldn't find any reference on the web. I guess the book is worth reading for students in CS and CPU/system designers and all those strange people dealing with computer systems design preferring to think of the things happening not taking ads as definitive source of information. Also it's a good piece of information for those interested in technology/computer history, analyzis and forecasting.