I’m reading The Soul of a New Machine, which I discovered by accident when looking for some other books. It’s a fantastic book! It details the development of the MV/8000 series of superminicomputers by Data General in the 70s/80s – their response to DEC’s VAX.
This book ties in with my own recent work and reading in several ways, which is probably why I’m enjoying it so much.
For my PhD research, I’ve been reading various academic papers going back to the 60s on the topics of automatic memory caching and virtual memory, and the variety of ways that problems have been addressed (hur). This book goes into surprising amounts of detail about the design and implementation of computer architecture, and to read Steve Wallach’s memory segmentation “Golden moment” explained in this narrative history was a delight for me. The motivation and concepts are quite different, but it brought to mind the explanations that can be found in Neal Stephenson’s Cryptonomicon (and even Cory Doctorow’s Little Brother) – an attempt to teach some computing concepts within the scope of a story.
My reading for research has brought to my attention the Burroughs large systems, computers produced by Burroughs Corporation in the 1960s – particularly the B5500. I have been fascinated to read how the company’s technology has persisted over time through mergers and renamings and many generations, to the extent that the MCP (Master Control Program) developed by Burroughs back in 1961 has some (vague) support still available in hideously expensive mainframes made by Unisys (including some kind of object-level compatibility, although I’m not sure how far back that goes).
Another reason that this book has been of interest is that it is a story of intra-corporate competition (and conflict) that seems to resonate with the stories found in The Inventor’s Dilemma – companies that grew quickly based on a good product, but that didn’t change as the market required. Data General spent a lot of time and money fighting to win in (what would prove to be) a superceded product domain, and while they still sold a lot of machines (from what I read, billions of dollars worth of the MV/8000 series), they perhaps changed too slowly and circumstances were against them. See more on the wikipedia page.
(I should say that I didn’t finish The Inventor’s Dilemma, and I haven’t yet finished The Soul of a New Machine – I’m aware that the first references the idea attributed to Tom West that the internal design of the VAX reflected the internal corporate structure of DEC, but I do not recall reading that part of Inventor’s Dilemma… I guess I should request it again)
At times I wish I knew more about computer history, particularly in terms of how hardware design has changed over time – there is a lot to learn from the past, I think, when it comes to implementing computer solutions now. For example, algorithms that made sense for managing or controlling dataflow between memory and disk are now often applicable between the processor and the memory, such has the relative speed of each changed. This is something that I have found to be useful in my research, having implemented a caching system that has more in common with RAM-to-disk virtual memory than traditional cache-to-RAM.
In fact I’d like to see taught (by which I mean “I’d like to teach”) a full-year program-your-way-through-computer-history unit at university :)
With regard to my reading about this book, I found this page where it was fascinating to me to see a post describing the backwards compatibility of the MV/8000 by Michael Meissner on usenet from 1987 – an engineer working for Data General at that time, and a name I recognise as currently working for IBM, making contributions to GCC in the areas of __ea support for the Cell BE SPU, as well as Power7 support. I also note that DJ Delorie, another name I recognise from the GCC mailing lists, worked at Data General (as mentioned on the DG wikipedia page). I am continually excited to follow threads connecting computing companies and people to different places and times. (Going back to Burroughs, this story of Don Knuth writing an ALGOL-58 compiler over a summer is quite an entertaining read. Also, I’d not noticed this interview before).
For further reading, I’m hoping to get a copy of The Race for a New Game Machine – about the development of the PS3 & Xbox360 processors, as well as Racing the Beam, on programming for the Atari VCS.