Click a photo to read one of the exclusive interviews
Craig Vaughan's "Lost" interview with John Mathieson
The following is an unpublished extract from Craig's Retrogamer issue 8 feature about The Konix Multi-system. I'm offering it in this section to make it more accessible as it's already in the full document offered as a download in the Contributions section of this website.
It is all Craig's work, and I am taking no credit for it - it reamins the copyright of Craig Vaughan.
John Mathieson designed the 'Flare One' chipset, which became the beating heart of the Konix Multi System. As he recalls, the prototype hardware started off life as an Amiga wannabe and ended purring inside the Atari Jaguar.
Even by today's standards, John Mathieson had a colourful start to his
computing career: "I joined Sinclair Research at the end of 1981, a graduate
fresh from Cambridge. At the time, the company was flying high. The ZX81 was
selling like hot potatoes, and the Spectrum was a huge breadboard on the
fourth floor of a tiny office opposite King's College in Cambridge. My job
was to test out the Spectrum's BASIC interpreter and find bugs in it. I went
on from there to being engineering liaison for the Spectrum launch,
travelling all over the world including going to Japan with Margaret Thatcher
to give a Spectrum to the Japanese Prime Minister. I went on to become
responsible for the hardware development of the machine".
With the Spectrum on shop shelves and proving to be a great British success story, Sinclair Research looked to be heading away from making home computers in a bid to dominate other potentially profitable markets. Mathieson relates: "While Clive went off into the business computer world with the QL, and later into the world of electric cars, there remained a core group of people within Sinclair who wanted to continue selling cheap home computers with dual educational and game playing capabilities - machines that the kids would want and that parents would be willing to pay for. We were impressed when the Amiga came out because as a computer it seemed to fit that role perfectly. But, we believed that the Amiga was over engineered - it could animate with sprites and with a blitter, but why have both? We figured that if you could implement a system that could animate the screen then you only needed one set of hardware to do it. So we proposed a computer called 'Loki' as a Spectrum successor. The idea was never more than a paper design because Sinclair sold out and Amstrad took over before anything productive could happen to further the project".
The idea stayed with Mathieson until 1986, when he made a decisive move: "Myself, Martin Brennan and Ben Cheese decided to go it alone to continue the idea. We formed a company called 'Flare Technology" and intended to fund it by doing contract design work whilst developing a prototype that would stun the world. On the way to 'Flare One' we designed a Spectrum clone for a Spanish company and through this met Alan Sugar. He wasn't impressed by the fact he was losing sales to our machine, but out of this brief legal entanglement we ended up doing work for him too, designing a hard disk controller for his early PC clones. Over time, 'Flare One' slowly came to life. By 1988 we had a working prototype and some cool demos running on it. We had a rotating 3D cube using a colour hold mode, some crude video capture demos from the Blues Brothers, and some other cool animation stuff. We received a lot of favourable press coverage, and worked our butts off trying to find someone to take the idea to production. The key to the technology was twofold: a blitter that could move pixels around, being limited only by how fast the memory ran and a DSP to generate synthesized sounds of previously unheard quality. The blitter idea was not new, but ours was far more flexible and game-oriented than earlier blitters. The DSP as this price level was completely new".
Having stunning prototype hardware was all very well, but Flare knew they needed a backer to further their plans: "Our first meeting with Wyn Holloway of Konix was at the Sheraton Skyline in Heathrow, during August of 1988. He wanted our technology to go into a project he had nicknamed 'Slipstream' - the famous games controller that looked revolutionary. The plan was to launch it in January of 1989. We had to combine the four simple ASICS of 'Flare One' into one LSI Logic chip, change from an 8-bit Z80 CPU to 8/16-bit 8088, and have all this ready to ship in 5 months! The excitement was tremendous. Wyn said that he would pay us a royalty of about a pound on every unit sold, and talked about sales in the tens of millions. He would pay us cash up front to start development, and could not wait to get started. Wyn is a great salesman and had us sold on the whole idea. His credibility was strong - Konix had a great name in the joystick business and it seemed like they could pull it off. We put our heads down and got to work. 'The Flare One' design was tweaked a little based on our experience, and some data paths widened to 16bits. By December we had working silicon, but it wasn't fast enough. We did another spin to move up to a full 16-bit 8086 and to integrate a floppy disk controller. By using a custom disk format we even managed to get 880K on a normal 720K floppy disk".
As work continued the technical specification of the machine increased in proportion to the hype generated by a salivating British press: "The graphics technology was pretty powerful too. We added 16bit pixel support; and used a new type of DRAM called Pseudo-Static that was actually DRAM but had most of the speed of SRAM. The competition at the time, the Super Nintendo and Sega Genesis, were sprite based, so our far more flexible and general blitter solution allowed the game writers tremendous flexibility in what they could do on stream, and the demos bore this out. Time slipped on and the January target was missed, In February Wyn was talking about deals with Lucasfilm and the volumes that we would generate, and we continued to anticipate all of those royalties! The revised 'Slipstream 2' silicon came back in July, and by September we were actually debugging production boards. Wyn had a software developer's conference in a hotel in Wales, which took place at the height of the project's euphoria: we proudly showed off our technology all day, then stayed up drinking with Wyn all night".
But, just as success seemed inevitable, somehow defeat was snatched from the jaws of triumph: "After that the project slowly fizzled out. We were paid for our development work, but Konix ran out of money. When they failed to turn up at the trade show they were supposed to be launching the system at we knew the project had died. Although matters fizzled on for a while, nothing of substance ever happened - very much a case of so near and yet so far. Chastened by this experience we decided next time we needed a larger partner to work with us on 'Flare Two'. We had been planning the technology for 'Flare Two' for some time, learning from the 'Flare One'/'Slipstream' project. We wanted to add 3D rendering for the first time to games and we wanted to do our own RISC CPU to provide the necessary power. 'Flare One' had demonstrated to us that we knew how to produce the technology, but it also proved that we needed a real partner from the beginning to fund it so we could focus on the design. Ben moved on, but Martin and I struck a deal with Atari, code named 'Project Jaguar', and the rest, as they say, is history"