What do you think?
Rate this book
Planning a Computer System: Project Stretch
322 pages, Hardcover
Published January 1, 1962
40 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 of 1 review
October 5, 2019
Info: Planning a Computer System: Project Stretch is an excellent example of (1950s) computer systems design, analysis, realization, and deployment. Editor Werner Buchholz coordinates this technical description of the IBM 7030 STRETCH. Set against the hugely successful but moderately powered IBM 650 and the expensive but powerful IBM 704, Project STRETCH aimed to achieve 100x better performance and to accommodate problems with much larger memory requirements, with an affordable, general-purpose computer. For this to happen, the STRETCH team would have to innovate across the board, from computer science to manufacturing.
+++ The team of designers is beyond amazing. Werner Buchholz, a Holocaust survivor, was involved in the teams of IBM 701 (a 704 precursor) and STRETCH, coined the term 'byte' (1956), and in 1990 received the IEEE Computer Pioneer Award. Frederick P. Brooks, Jr., of later IBM System/360 renown. Gerrit A. Blaauw, the Dutch computer scientist, and co-author with Brooks not only of the System/360, but also of the extraordinary textbook Computer architecture: concepts and evolution. Robert W. Bemer made one of the first publications on time-sharing (in 1957) and went on to become 'The Father of ASCII'. Erich Bloch, another Holocaust survivor, served as Director of the US National Science Foundation. John Cocke went on to invent the basic concepts underpinning the RISC architecture, co-designed the IBM 801 minicomputer, and was awarded both the ACM Turing Award and IEEE John von Neumann Medal. Edgar F. Codd went on to invent relational databases at IBM in the late-1960s; at SIGMOD 2019, the WiFi password was still a play on 'In Codd we trust'. etc.
++ The book covers: the architectural philosophy of the system, the overall system design, detailed design and analysis of the instruction set architecture (ISA: data formats, instructions, instruction sequencing, data indexing, IO instructions) and organization (the CPU, IO control and operation, multiprogramming, the look-ahead unit).
++ There is additional material on the IBM 7950 Data Processing System, which focuses on processing non-(arithmetic-and-logic) data. The chapter also includes Buchholz' definition of the byte as a basic 8-bit unit of information.
+ There are also two appendices, on summary data and on programming examples. The former is a rather simplified technical reference of the ISA, plus a list of IBM computers and their characteristics. The latter is a primer about programming the STRETCH system, through a few examples, both of scientific computing (oh, the joys of matrix multiplication!) and of non-arithmetical processing (one example, of editing a message input from a typewriter).
+/-- Some of the claims to fame: first transistorized supercomputer, first multi-programming/time-sharing environment. Warning: Some of these claims, have not been validated by historians of computing; for example, that about the design of multiprogramming is not publicly documented elsewhere, and there is a still active debate about who invented this -- Brooks claims here a design decision on multiprogramming made in 1956, Bob Bemer claimed later a first public presentation on multiprogramming and pointed to his magazine publication in 1957, but the concept seems to have appeared earlier, Feynman mentions a type of (manual) time-sharing by the Los Alamos computing team in Project Manhattan, the concept seems to have been deployed by the NSA computers ROGUE (designed 1954, prototyped 1955, operational 1956) and ROB ROY (designed Feb 1957, before Bemer's article, operational Feb 1960, before STRETCH in May 1961), the idea was explicitly mentioned in publications 1955 by Rochester and in 1956 by Friend, and the concept itself was popularized by a publication of Strachey in 1959 (in an audience that included STRETCH team-member Codd, who claimed on the spot that IBM had an earlier claim to what Strachey was presenting).
+++ The team of designers is beyond amazing. Werner Buchholz, a Holocaust survivor, was involved in the teams of IBM 701 (a 704 precursor) and STRETCH, coined the term 'byte' (1956), and in 1990 received the IEEE Computer Pioneer Award. Frederick P. Brooks, Jr., of later IBM System/360 renown. Gerrit A. Blaauw, the Dutch computer scientist, and co-author with Brooks not only of the System/360, but also of the extraordinary textbook Computer architecture: concepts and evolution. Robert W. Bemer made one of the first publications on time-sharing (in 1957) and went on to become 'The Father of ASCII'. Erich Bloch, another Holocaust survivor, served as Director of the US National Science Foundation. John Cocke went on to invent the basic concepts underpinning the RISC architecture, co-designed the IBM 801 minicomputer, and was awarded both the ACM Turing Award and IEEE John von Neumann Medal. Edgar F. Codd went on to invent relational databases at IBM in the late-1960s; at SIGMOD 2019, the WiFi password was still a play on 'In Codd we trust'. etc.
++ The book covers: the architectural philosophy of the system, the overall system design, detailed design and analysis of the instruction set architecture (ISA: data formats, instructions, instruction sequencing, data indexing, IO instructions) and organization (the CPU, IO control and operation, multiprogramming, the look-ahead unit).
++ There is additional material on the IBM 7950 Data Processing System, which focuses on processing non-(arithmetic-and-logic) data. The chapter also includes Buchholz' definition of the byte as a basic 8-bit unit of information.
+ There are also two appendices, on summary data and on programming examples. The former is a rather simplified technical reference of the ISA, plus a list of IBM computers and their characteristics. The latter is a primer about programming the STRETCH system, through a few examples, both of scientific computing (oh, the joys of matrix multiplication!) and of non-arithmetical processing (one example, of editing a message input from a typewriter).
+/-- Some of the claims to fame: first transistorized supercomputer, first multi-programming/time-sharing environment. Warning: Some of these claims, have not been validated by historians of computing; for example, that about the design of multiprogramming is not publicly documented elsewhere, and there is a still active debate about who invented this -- Brooks claims here a design decision on multiprogramming made in 1956, Bob Bemer claimed later a first public presentation on multiprogramming and pointed to his magazine publication in 1957, but the concept seems to have appeared earlier, Feynman mentions a type of (manual) time-sharing by the Los Alamos computing team in Project Manhattan, the concept seems to have been deployed by the NSA computers ROGUE (designed 1954, prototyped 1955, operational 1956) and ROB ROY (designed Feb 1957, before Bemer's article, operational Feb 1960, before STRETCH in May 1961), the idea was explicitly mentioned in publications 1955 by Rochester and in 1956 by Friend, and the concept itself was popularized by a publication of Strachey in 1959 (in an audience that included STRETCH team-member Codd, who claimed on the spot that IBM had an earlier claim to what Strachey was presenting).
Displaying 1 of 1 review