The Roots of UNIX for the IBM Mainframe (part 3)

In 1975 Ken Thompson had taken a sabbatical at U.C. Berkeley and catalyzed them into becoming the west coast center of UNIX expertise. While there, he became friends with Denny Koch.  Denny went on to get a job with Amdahl Corp. Amdahl  was a hot Silicon Valley startup taking on the ogre named IBM, and they had an excellent college recruiting program which got lots of great people from great places.

At the time, Amdahl had little or no software products, but they needed a lot of software people for their design automation group. Denny and others like him who had used UNIX in school started agitating for a better time-sharing software development system.

So I’m sitting in the UNIX room at Bell Labs in August of 1977. Ken Thompson is also there and he gets a call from John Hiles of Amdahl asking him if he knows anything about the rumors that there is a UNIX port for the 370.  Ken says, “Hold on, you need to talk to Tom” and hands me the phone!

I explained to John the state of the port and he strongly hinted that they’d be interested in hiring me. Alas, I had another year of school left.  But in late January of 1978 I finally made it out to Sunnyvale, CA for a job interview.  The weather was fabulous but I was still wearing the snow boots I needed to escape New Jersey!

I visited Amdahl again during spring break for a week of consulting.  By then, John had gotten a proper UNIX license from Bell Labs and permission to use the port we developed at Princeton.  I carried UNIX/370 to Amdahl on 4 DECtapes (pictured). The DECtapes weren’t much use to a mainframe, so John and I rented some PDP-11 time from the local Digital office. I was able to boot UNIX there and then copy the DECtapes to a 9 track tape.

A few Amdahl folks like Andy Tucci and Dennis Andrews started messing with the code. Then in June I  joined Amdahl full time and there were a bunch of other new grads put on the project. I shared an office with Mark Linton (also Princeton ’78) and we got a lot done in a short time. At first John Hiles managed the project, and then Larry Whitlock after John was promoted.

By early 1979 was had a full Version 6 UNIX system running on the Amdahl 470V/6 under VM/370. We named our system ‘Au’, pronounced ‘gold’. We had real and happy users from the design automation group. I’m especially proud of the driver I wrote for the IBM 3270 terminals which made them usable with UNIX.  Full-duplex and type-ahead were completely unknown in the mainframe world, but I managed a very close emulation.

Later in 1979 Amdahl received the Version 7 UNIX tape from Bell Labs, and in 1980 announced Amdahl UTS would be available to all comers. But I started getting more involved with data communications and less with UNIX. Until a call from Scott McNealy…