Even though I only scanned the article, many of those computers appear in Turing's cathedral G.Dyson[1], book which I enjoyed, given that there is not a lot of history lessons at the universities (at least the one I went to in Buenos Aires... UBA).
Might be expensive to house and maintain something like the SSEM or the CSIRAC, but universities have spent money on sillier things.
I love getting my Dad talking about some of the labs he's worked in over his career and kind of wish I could get a sense of how much has changed over such a short time.
After a couple go-rounds with that, we were then allowed to use the terminals, mostly Visual 200, with a few VT100s, an LA36, and an LA120. The Visual 200 wasn't a particularly good terminal; it seems there was always at least one out of order at any given time.
It's kind of scary that a $5 Raspberry Pi Zero is much faster and has vastly more RAM than the 11/34, which cost well north of $100K back then. Not to mention that there was a grand total of 28 MB of storage on that 11/34 in the form of two RK07 disk packs - you could fit thousands of RK07 images on a single 64 GB MicroSD card.
"You can buy FETs that have 1970s performance specs in 2017, very cheaply in fact, but in 2017 you can also buy FETs that output many watts of power at 10 GHz x-band microwave freqs. Here's a DEC flip-chip module containing two flipflops that topped out around 30 or so MHz in 1965, your assignment is to use modern transistors in the lab to make a modern work-alike operate over 10 GHz"
> buy FETs that output many watts of power at 10 GHz x-band microwave freqs.
Part number? Gate drive for that must be 'interesting'.
That said, on emulation, it's a great idea. It's easy to forget how many of the assumptions of C-like languages and Unix-like operating systems are built into the hardware, particularly when there are other ways to go about the task. There are reasons the industry has gone the way it has, but it's worth keeping those ideas around for posterity.