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.
