> There's not a lot of "off the shelf" in terms of mainframes
Not true-there’s still a huge amount of off-the-shelf software for IBM mainframes, and in the 1980s there was a lot more than there is now - see e.g. https://lookupmainframesoftware.com/ - in particular, the early-to-mid 1980s and earlier were really the heyday of scientific and engineering software on IBM mainframes (e.g. the original mainframe-based versions of CATIA and CADAM), before it migrated to Unix workstations. From Ken’s comment above, I know Intel was actually using Amdahl UTS, a mainframe Unix port, which is a bit of a midpoint in that transition. And of course, someone like Intel (both then and now) is going to use a lot of custom software in CPU development, whatever off-the-shelf bits they might be able to leverage.
Not true-there’s still a huge amount of off-the-shelf software for IBM mainframes, and in the 1980s there was a lot more than there is now - see e.g. https://lookupmainframesoftware.com/ - in particular, the early-to-mid 1980s and earlier were really the heyday of scientific and engineering software on IBM mainframes (e.g. the original mainframe-based versions of CATIA and CADAM), before it migrated to Unix workstations. From Ken’s comment above, I know Intel was actually using Amdahl UTS, a mainframe Unix port, which is a bit of a midpoint in that transition. And of course, someone like Intel (both then and now) is going to use a lot of custom software in CPU development, whatever off-the-shelf bits they might be able to leverage.