The Rise and Fall of Silicon Graphics

I believe nVidia was started with a lot of SGI's core technology ; not "I have a good idea and I can't do it here" and more like "let me just take this stuff I doubt anyone will notice". I think SGI sued but didn't really pursue the matter because they didn't really see nVidia as a threat. I think Jensen was pivotal in this "technology transfer".

Regarding computing cycles, boom/bust, I recently re-read Soul of New Machine and was struck by how much the world has NOT changed. Sure we're not talking about micro/mini-computers and writing micro-coded assembly but the whole "the market is pivoting and we need to ride this wave" and "work like a dog to meet some almost unobtainum goal" seems to still underpin being an engineer in "tech" today.

From my reading SGI was already dead and falling apart by that time. If you look at 3D, SGI had two graphics architectures in the 90s: RealityEngine from 1992 and InfiniteReality from 1996. They never managed to release a follow-up to IR. Similarly everything that came after about 1996-97 was a refresh of a prior product with only marginal changes. And then they went bankrupt in the early 2000s. So SGI had really only a very brief productive period that was over by the second half of the 1990s.

SGI also never had a presence in business critical applications which gave some of the other vendors more momentum (HP-UX/PA-RISC, VMS/Alpha, Solaris/SPARC).

Well,

Most Hollywood effects were all done on SGI systems before the slow migration to Linux. Renderman, Maya, were all SGI first-party programs.

Also SGI made huge advances in NUMA and machines with dozens of CPUs/processors before most other companies ventured into this space.

But not business critical like IBM CICS or Java.

1. https://en.wikipedia.org/wiki/NUMAlink

2. https://www.cs.ucr.edu/~bhuyan/CS213/2004/numalink.pdf

3. https://cseweb.ucsd.edu/classes/fa12/cse260-b/Lectures/Lec17...