Hacker News Reader

Comment by chasil | Hacker News Reader

DEC spent an enormous amount of money on an emitter-coupled logic (ECL) bipolar implementation as the VAX 9000.

They came to discover that the MicroVAX hit ~70% of the performance at a fraction of the cost.

That was an expensive dead end.

"Production problems pushed back its release, by which time these fears had come true and newer microprocessors like DEC's own NVAX offered a significant fraction of the 9000's performance for a tiny fraction of the price.

"Roughly four dozen systems were delivered before production was discontinued, a massive failure. "

https://en.m.wikipedia.org/wiki/VAX_9000

View on HN

We had one of these at CERN in the experimental physics group I was in 1990-1991. I had no idea they were rare or that the line was a failure. It was certainly faster than the other machines the group had access to at the time (except an onsite Cray, access to which was restricted only to members from "Western" countries for political reasons).

Our shop floor system is still written in VAX ACMS, using TDMS for menus, and Oracle Rdb for SQL.

I still have an account. It runs on the Charon emulator.

A little after your time, I wrote code on a DECstation 5000/240 running MIPS/Ultrix. I'm surprised that wasn't faster, and/or available to you.

The vector extensions might have been your niche though.

It wasn’t a discovery: DEC knew the 9000 wouldn’t be competitive against VLSI implementations! DEC’s board really should’ve replaced Ken Olsen by 1985 at the latest.

> There are several quotes by prominent engineers on the NVAX project that describe Olsen's unwillingness to kill the 9000 even after being told point-blank that it would not be competitive by the early 1990s,[14] and his outright rejection that such a thing was even possible.[18]

At Sun we were delighted that DEC's board wouldn't replace Ken Olsen!

The line between stubbornness and persistence is very thin. It's easy to rationalize post-factum that A was right to push his idea even when everyone told him it won't work, and B was stupid not to listen when everyone told him it won't work.

DEC tried something and failed, that gives me more joy than seeing Intel failing without trying. Transition from Otellini to Krzanich driven by the risk-averse board was painful to watch from within.

Right; if you claim Olsen should have listened to all the experts in 01989 telling him VAX was doomed because it was CISC instead of RISC, you are also pretty much claiming Andy Grove should have listened to all the experts in 01989 telling him the 80486 was doomed because it was CISC instead of RISC. They were right about the VAX and wrong about the 80486, but that wasn't clear for about five years. See my comment at https://news.ycombinator.com/item?id=43811508 for more details.

ECL vs. CMOS is a slightly different battle, but HPC companies (and Amdahl) kept shipping new commercially viable ECL processors until the mid-90s, and the VAX 9000 was sort of aimed at those markets. Tera shipped their first bipolar MTA in 01998, and if they'd managed to ship it a year or two earlier, it might have been commercially viable too. So it wasn't crazy to think that an ECL VAX 9000 would be the success that the ECL VAX 8000 series had been. It was just wrong.

Supposedly, DEC started to lose its way when Olsen no longer had Doriot around to mentor him.

Olsen was an old-school MIT guy - an engineer making "affordable" tools for other engineers and scientists. It was a personal crusade for him, and his biggest competitor was always the ghost of the IBM System 360.

He literally couldn't imagine the economies of scale that VLSI made possible, or how to pivot from DEC's traditional customer base into commodity computing at one extreme and the more open Sun/SGI workstation market at the other.

There's a parallel universe where DEC invented the commodity PC and made the Internet run on it. It's likely quite an interesting place.

DEC always had top-tier engineering talent and had the Alta Vista search engine well before they understood how valuable it could be.

DEC's management and vision were also top-tier, until suddenly they weren't.

Reply on news.ycombinator.com