This strange machine was shown at Comdex in 1987. It was designed for graphics intensive applications taking advantage of the parallel processing capacities of the INMOS transputer.
Two models were sold: one of them was a card which could be connected onto a Mega STf bus expansion to use its peripherals (keyboard, disk, etc.). The other was a computer in which there was a motherboard with a single transputer plus a card which contained all the Atari Mega STf hardware to use its peripherals!
The motherboard included slots to add up to four more processor cards, each
with up to four transputers, to enable multiprocessing.
It ran under Helios, an operating system done by the British company Perihelion, derived from UNIX but specially adapted for parallel processing).
A C language was adapted and a parallel processing language called OCCAM. Only few of these computers were sold, most of them to Kodak.
I managed to get an ATW shortly after the project was abandoned and because of the lacking documentation contacted Perihelion to see if I could get anything from them. During the conversation it was clear they were really pissed. They said they has a deal with Atari. Atari would deliver the hardware and do the distribution, Perihelion would deliver the software. With the ATW was a bundle of documentation including registration and guarantee cards. On reception, Atari would forward the registration cards to Perihelion. It turned out they didn''t. This was either the start of a nasty relationship, or the last straw, but the conversation made it quite clear that this was one of the major issues that broke the cooperation.
I still find it a great loss, the whole thing - including the T800 - was lightyears ahead of its time.
Tuesday 27th December 2011
Franz (Netherlands)
There were 2 reasons the Abaq/ATW failed, the price and Atari''s inability to market the thing, never mind market it to places that would actually make use of it.
In terms of processor "Power", Intel chips could not match the Transputers ability to do complex calculations until 5 years AFTER the ATW had been made and discontinued, and could not do multi processor calculations as well until around 2007 (Some may argue it still can''t do this as well, even with multi-core processors but that is another argument and I''m not siding either way). Bare in mind that up until 2004, most servers could not handle more than 4 CPU''s as adding more than this started to reduce performance, whereas in 1987 you could farm many more transputers which would work in harmony AND each individual chip could process calculations faster than any chip in PC''s of the time. The failure of the ATW lies at Atari''s door. They were trying to make technology for markets they had no idea about or how to approach.
It was a brave and bold move at the time, but was too much for such a company, that is why it failed.
ShareThis
