Hacker News Reader

Comment by unixhero | Hacker News Reader

Finally some acknowledgement for Norsk Data.

It was destroyed by a right wing government which refused to "pick winners", and grant broader governmental innovation funding. And of course in addition not winning the CPU arms race which happened at the time.

However to think that Norway had an infant computer industry, with its own NON-UNIX mainframe system, complete with hardware in several sizes, operating systems and so on. It was even cloned by the Soviet Union, and cloned spare parts were sold behind the iron curtain!

What this company achieved was monumental.

View on HN

There are more examples. Between the 60ies and early 90ies, Philips in the Netherlands sold their own line of proprietary computer systems such as the P800 series (PDP/11-like) and P4000 series (a Cobol programmed office computer). The Philips computer division was later sold to DEC. Information about much of this hardware is nearly nonexistent.

Indeed. There were over two dozen minicomputer manufacturers at the peak in the early 1970s in just the USA. I don't know off the top of my head, but I imagine there were just as many in Europe.

Most of the machines and their software are entirely lost to history at this point, with the companies bankrupt or dissolved and documentation and hardware all scrapped.

Philips also had MSX based computers as well

-Additionally (while on the subject of computer history and norsemen), the first ARPANET node outside the US was located in Norway; NORSAR (The Norwegian seismic survey) was hooked up in June 1973 - presumably the seismic activity of interest to the US was Soviet nuclear tests.

This is funny because Norsk Data computers were used in VNIIPAS ( All Union Scientific Research Institute for Applied Computerized Systems) for first Soviet X.25 network, connected to the rest of the world via Austria.

ND also had a Unix System V port running on their machines (the ND-4000 but not ND-800 series if I recall correctly).

Source: worked for ND 1988/9 (not in Norway, though) when the end was nigh and they were looking for a plan B as customers moved to open systems (= Unix)

Edit: didn't recall correctly; was ND-570 not ND-800 and ND-5000 not ND-4000 (in late 1980s these series were basically the same at the CMOS level and an ND-570 could be on-site migrated to 5000 by merely cutting a wire on the board with a knipper to engage additional instructions or address lines)

Wikipedia comments that the ND500/ND5000 architecture wasn't especially Unix friendly as it was based on a 16-bit front-end & I/O ND100 processor to the 32-bit ND500/ND5000 cpus. Also that this was ultimately a system bottleneck for competition in general.

Yeah, key to programming the ND500 and ND5000 series was to utilize the ND100 properly. Way too much software would overload the ND100, while leaving the ND500(0) running idle.

The ND-100 was a bottleneck for the 500 series, and initially also for the 5000 series. But as time went by the 5000 series acquired intelligent I/O interfaces sitting directly on the Multi-Function bus on the 5000, and there was less and less of a bottleneck. The systems I worked on had to get the absolute last possible drop of I/O performance out of it, and by having the ND-100 do its own I/O in parallel with I/O on the 5000 the whole thing worked very efficiently in the early nineties. The bottleneck, near the end, moved to the 5000 (or 5700, as it were) CPU instead. A Sun 4 could run circles around the 5700 in processing time - at one point I ran all our Fortran code through f2c and executed it on the Sun. Wow. That's when I started to talk with the customer about a possible rehost (which we did, to an SGI).

I suspect that's not entirely accurate. The ND-5000 series was a real upgrade to the ND-500 series, with distinct silicon.

However, I have seen the cutting-a-wire-on-a-board in real life. We bough an upgrade from an ND-530 to an ND-570, and a technician came with a knipper. It was surreal: The upgrade was north of 100KUSD. And all they did was cut a wire.

It's what made the rounds in dev circles where I worked. The technician would have still to upgrade/re-install the OS, and to prevent the perceived rip-off with the knipper to be noticed and make the customer feel bad, I think they also discussed that the technician should do some theater job such as walking in with a packaged CPU upgrade kit so that the upgrade took at least 2 hours. It was not a rip-off, as the 5000 had, like all CPUs, a costly upfront development investment. Idk but the reason why some ND-500 -> 5000 upgrades may have come with real silicon while others only need the knippers job is maybe that once the 5000 CPU was produced in quantities, the 500 that were still sold were equipped with 5000 CPUs from that point on, to save an expensive CMOS production line. Even IBM, by around 1990 at the latest, produced all their non-X86 CPUs (for P-Series aka RS/6000, I-Series aka AS/400, and Z-Series aka S/390) off the same silicon.

Agreed. I recall rescuing a 5000 series CPU board from some systems being retired. The boards had originally cost ~50K GBP each and consisted of four seperate 12" x 16" boards bolted together in a 3" thick stack. The boards had several MCUs including M68k plus lots of 74series logic chips (numbers approximate - it was a long time ago).

The 5000 series was different from the 500 series, it was a new generation of computers and the 570 wasn't a 5000 in hiding.

But it's basically correct that you could field-upgrade some of the models to a faster model, not exactly by just snipping a wire, but you could e.g. change the cycle time, and enable caching. But to go from the 5200 to the 5400 you would need to change some boards too. Here's an overview: http://www.ndwiki.org/wiki/ND-5000_family#System_performance

> It was destroyed by a right wing government which refused to "pick winners", and grant broader governmental innovation funding.

The Conservative party took over the Norwegian government in 1981 and held the power alone or with coalition parties until 1990. I am by no means an expert in Norsk Data trivia, but if Norsk Data was the second largest company in Norway in 1987, I have a hard time understanding that Norsk Data was destroyed by 'the right wing government which refused to "pick winners"'.

In this article 1] the government is not mentioned and the author Arild Haraldsen asks the question: "Could Norsk Data have been Saved? Nothing suggests that it could".

1] https://translate.google.com/translate?sl=auto&tl=en&u=https...

Of course it could have been saved with proper funding, restructured, could have been nationalised, many other options.

They also sold CAD workstations in late 1980s with relative success.

Reply on news.ycombinator.com