The mass extinction of Unix workstations

New Mexico Tech had Sun 3 workstations based on the 68k line, then we switched to Sun SPARC workstations which were small, powerful and really cool. (There was that time we had two workstations that each had two halves, one of which was anchored to the desk with a cable. We swapped the two halves and had a machine that was anchored twice and one that wasn't anchored at all.)

Just as I was graduating, Linux had come out and the word was a 486-based machine would beat the pants off a Sun at a lower price. I went to grad school at Cornell and built an AMD 486 machine that I named after this thing

https://safebooru.donmai.us/posts/3296603

and overclocked. I was in the habit of building the kernel each time a new one came out and I wrote to LKML about how I had cc crash a lot when I was building the kernel and Alan Cox said "don't overclock, it will burn out your machine". I went back to the standard clock and the compilations were OK but the CPU failed for good a few months later.

Cornell had a relationship with IBM and we had a lot of PowerPC machines running AIX (we had an IBM SP/2 supercomputer cluster based on them) but around the time Windows 95 came out we got a grant from Intel and Microsoft to get a bunch of x86 machines. The plan was to run Windows on them but very few grad students wanted to do scientific computing, at first we had maybe 4 Linux computers and 12 computers running NT 4. The two people who used the NT machines was a guy who loved Windows and me who hated Windows (and was seen as the leader of the opposition) but I would use vnc to log into a Linux machine and not have to fight for one.

We had our sysadmin rage-quit one day and he went to be the king of printers at Central IT which was still using Unix. The professor who knew Bill Gates threw up his hands and switched most of the machines over the Linux.

There was one guy in the chemistry department who bought a stylish purple SGI workstation without enough RAM. He couldn't get much done with it but he left it plugged into the Ethernet for months until we realized that the root password on the machine was the empty string.

I had both the 32bit sparc and UltraSparc machines running Solaris, 21064 Alpha running Linux, 21264 Alpha running Tru64, and also use of pa-risc boxes with HPUX, MIPS running IRIX, and Power running AIX. All of it was second hand old stuff by the time I got it, of course. Fun times.