What a DEC-20 Cost in 1976 (2007)
dec20.blogspot.com
dec20.blogspot.com
We also had APL for a numerical methods class on that system. Yes, we had some terminals with the special APL symbols; I think they were LA-36's.
Boyer and Moore put in an actual integer representation after that. We got our own VAX at Ford Aerospace and ported the prover to that. Then we could get some work done on proof of correctness.
EDIT: The post might be something like: "Why we bought a DEC for 300K and never thought twice about it"
Input customer orders, schedule production, track inventory, print shipping labels, do accounting, send invoices, and type memos... all in one integrated system.
And if you had multiple offices, they could each have one and they could exchange data throughout the day.
It was truly a revolution in productivity.
Obviously, you could run Empire on it!
No idea how much those cost. I was happy we didn't drop it, because serious injury was quite possible.
I often wonder what happened to that Vax. For a couple of years, I was the only person using it. It was glorious.
Wouldn't that be more like 8 years later? 1986 for the 12MHz release (originally targeted at 16), add at least a year to get up in speed and out in people's hands. And didn't the VAX have a much faster disk system, and few i386's tape backup?
BTW, how reliable were your RM80s?
Our RM80's were trouble free from what I can remember. We had an air conditioned, electrostatic filtered clean room for our systems which also included some PDP11's and PDP11/40's with RL02's and RK07 disk drives. The room had very limited access and you had to wear booties. The only time we lost a drive is when one of the RK07's went crazy, slamming the heads against the hard stops causing the heads to crash onto the disk.
I asked because the RA81 which was released in 1982 was, in MIT's Project Athena experience, awful, they had some sort of problem with glue, and they were losing 1 a week out of a population of no more than 40 drives (!). But it sounds like they were more careful, more lucky, whatever with the earlier RM80s, which I don't remember hearing anything bad about.
Everything went well until the RM05 (256MB) disk pack performed a random-seek confidence test; it was then that we discovered that its feet had not been screwed down to the floor and the whole unit started to breakdance around the computer room like a deranged washing machine. Two of us leapt at the unit and held it down until the test completed.
As for the Sun, using SCSI disks and a decent OS more or less compensated for limitations similar to Dell's circa 2000 PC.
Totally killed by a modern $5 Pi Zero.
If you took a Pi back in time no one would know what to do with it. Linux would be recognisable, Ethernet would be recognised by Xerox Parc and a few experts, and only at 10 MB/s - but USB, HDMI, SC cards, the SoC, and the web would all look like advanced alien technology.
A 2040 was a PDP-10 with a 30 MHz KL processor running around 30 MIPS and TOPS-20 (the source of the "20" in the name) a quarter century ago.
30e6 * 60 * 24 * 365 * 25 = about 4 e 14 instructions.
I can't get a straight answer but hundreds of MIPS sounds reasonable for a pi zero.
4e14 / 500e6 / 60 / 24 / 365 = about a year and a half
So sometime in 2017 a pi zero would finally catch up to a DECSystem-2040 which had a quarter century head start. Of course the 2040 has been continuously producing intermediate results for the first 25 years of the "race" while the pi output no results at all having not being invented yet. Also after 27 or so years the pi will somewhat surpass the 2040.
Also the IO bandwidth and scalability of old mainframes usually puts newer hardware to shame. "My calculator has more power than a 1960s NASA mainframe that sent men to the moon" "Oh so you're saying you've been to Mars, then?"
On one hand, 4e14 instructions executed would have been enough to run entire civilizations, some years ago. On the other hand, its not enough to boot modern Windows, although despite the difficulty is none the less not accomplishing anything in the real world.
Compuserve in say 1983 was a DEC service bureau moonlighting at night as a very expensive per hour super BBS system. Like $5/hr at 300 baud. And yes it was a network of PDP-10s as per the linked article (although I seem to remember KI processors not KL's)
Why did you make me remember that!
We used compuserve to support our (retail, boxed Mac) software. Then we had to expand to AOL to cover "the outsiders". And then... Eternal September.
(Some day, "like us on Facebook" will be as obsolete as "AOL keyword BLAH", which you can still see at the ending credits of movies from the 90s. I don't recall a similar push for compuserve.)
I wouldn't be so sure. USB, for example, isn't that different from RS-232, it's just faster.
There were people in the 70s (I was one of them) looking to build a handheld tablet. We were defeated (obviously) by the needed display and miniaturization technology being a couple orders of magnitude (at least) inadequate. Our ideas were awfully crude by current tablet standards, but we would have recognized a modern tablet, just like I would have recognized my current desktop as not conceptually that different from an LSI-11 box with a terminal.
USB 3 runs at low microwave frequencies. There's a lot of engineering between that and RS232 running at a few kbps.
I had both some years later. Both equally out-of-date. The car actually fell apart somewhat catastrophically on the road in 1982, because no rust-proofing.