I thought wow: Digital ruled the internet, though the PDP-1 must surely be a typo. I was looking out for a PDP-8 though but I admit that entering a TCP/IP stack through the front panel switches would be a challenge.
Considering that internet research was funded by DARPA (the DEFENSE Advance Research Projects Agency), and was intended to connect non-classified military and research networks, the presence of the Pentagon, NSA, and Mitre (a huge mil. contractor) on that graph should be no surprise at all. If you thought the interwebs were dreamed up by a bunch of college hippies so that we could all surf porn effortlessly, you should really hit the refresh button on your historical context.
And today, youād almost call that a distributed state machine.
So many PDP machines back then. Plus a few CDC systems thrown in for good measure. Those CDC machines were in place well into the late 80s.
You were six years old and connected to the internet? good lord. Iām envious.
Many PDP-11s and early VAX models had core memory, semiconductor memory was available in 1977, but was very expensive. PDP-8s could run a timesharing system called TSS-8, with the memory maxed out to 32K (12 bit words!) it could run up to 32 users more or less concurrently, who could develop programs in assembly language, FORTRAN (stripped down), and various interpreted languages like BASIC. So, it was quite possible he was connecting to a PDP-8ā¦
No, not a typo, in 1977 I believe there were actually either 3 or 4 PDP-1s connected to ARPANET, the one shown on the map was located in the CS department at Harvard, MIT still had one in EE department (which is now at the Computer History Museum) and one in the Tech Model Railroad Club (you could control the trains via Telnet, IIRC), and BBN (the primary developer of ARPANET) had one. All of these PDP-1s were heavily modified, they were somewhat more capable than when they left the DEC factory in the early 60s, and were still occasionally used for research (or running model trains).
Ahhhā¦ I think you misunderstand me;
The surprise isā¦ that anyone should be surprised.
All sorts of machines that could never hold a TCP/IP stack nevertheless made it onto the ARPANET. Most of the heavy lifting was done by the IMPs to which the hosts were connected - they were more than just routers. Hosts themselves only had to run a relatively small Network Control Program (NCP), which did local bookkeeping on connections and shoveled the data to and from user programs.
1 Like
by the way, did berkeley really not have a machine on arpanet in 1977? i see lawrence livermore, and even sumac but no cal. or am i misreading this?
(who remembers FTPing to sumac-aim? in the 90s i used to walk past that machine room in the medical campus, which had a little display in the window with some history - i decided it was a geek World Heritage Site. i think itās gone for good now. )
Ah, another ITS user reads boingā¦ I was in high school, in '73, took the bus into town, and became DP@MIT-ML (none.of.those.newfangled.dotted.hostnames thank you, and an ASR-33 didnāt have lower case). I have had an email address ever since. (and I now have a piece of MIT-ML in the form of one of the blinkylights panels that topped one of the memory racks - I think it took 4 of them to get a full 256kw on a KI-10) The arpanet of the time of this map was just starting to transition to ip/tcp, there was a far simpler protocol initially. (the original protocol had a 256 host limit). On some machines, you essentially only had telnet access, with the IMP (a honeywell 316 at least initially) appearing as a hardwired terminal to the remote machine.
No Berkley wasnāt on arpanet - the closest geographically was likely the Stanford AI lab. They didnāt really start the big Unix development thing until the VAX hardware appeared. (they had be doing some work on the -11, but I donāt think they distributed things until they put it on Vaxen). Their lack of an arpa connection is likely part of the impetus for UUCP.
Oh yea, hardware - all the ātraditionalā 8ās were core memory machines. Most PDP-11ās were core as well, but you had the option of solid state in some of the later models. When the -8 and -11 got reduced to a chipset, those were solid state only. All vaxen were solid state memory, core was never an option. It took 4 boards each at least as big as a large pizza box, to get 1mb. The memory controller could cope with up to 4mb, and you could add a second controller for an almost unheard of 8mb.
Core was fascinating stuff, the rings on the later stuff were impossibly small (like coarse ground pepper), yet they never managed to make a machine to thread wire thru them, it was always done by hand. (each core had at least 3 wires fed thru it, an x and y address lead, and a sense wire. Some had a 4th wire providing addressing for a 3D layout). The contents would last more than a few months, but reading the contents was actually a ādestructiveā operation, you had to re-write any cell you read from.
Back when I made my living with embedded computers, I read data out of boards that had been sitting on a shelf for over a year. You could also use them somewhat like a PROM, loading up a board on one machine, then moving it to another to run from. (after all a disk drive was much too expensive and temperamental to put one on a factory floor.)
1 Like
Isnāt LBL in the upper left Lawrence Berkeley Laboratory? That and Lawrence Livermore Labs are affilliated with UC.
I remember being at the Indiana University Computing Center in 1974 and hearing about ARPANET; I thought they had a node there, but I donāt see it on this map.
oh yeah, they are affiliated with berkeley but they are not the berkeley CS department. i guess as rjnerd says the berkeley stuff (BSD, etc) did not get going until the widespread adoption of vax machines. by that time the arpanet/internet was a lot bigger than shown here in 1977.
This topic was automatically closed after 5 days. New replies are no longer allowed.
