Hennessy and Patterson's Computer Architecture: A Quantitative Approach was the first really good textbook I'd read. It was a proper advanced undergrad textbook. It covered things from first principles, with detailed, highly-readable explanations that went on to advanced topics. The references to the literature both demonstrated how this tome of most of a thousand pages was but an introduction to the field, but also showed how to get to the cutting edge of the field.
And one of the great things about it was its real quantitative approach. Design trade-offs were illustrated through benchmarks, applications of Ahmdahl's Law, etc. The excitement of the RISC revolution was palpable in the text. At the time, if you wanted a high-end workstation, you didn't get an x86 box, you got a computer with a RISC CPU. Five main architectures were used as illustrations throughout the book - the Alpha, PA-RISC, MIPS, Sparc and PowerPC. Now, only Sparc is holding on by its fingernails in the server market. We're in a strange world where x86 is still king.
When decorating my study, I wanted to produced something which harked back to the time when it briefly seemed that logic might beat marketing momentum. A little display of examples of each of the architectures (with chips taken from decently spec'd machines - no getting the MIPS out of a washing machine!). This was a few years ago. A bit of ebaying got me a selection of chips - along with some rather funky heat sinks. However, with children, I never found the time to assemble the display.
Finally, though, I have. Not particularly impressively - it's rather half-done, but as a parent, this is small victory for me! Ideally I'll find time to build it into something more case-like, but I'm surprisingly proud of myself!