In 1999 I helped build a machine that did what a Silicon Graphics supercomputer did, out of three ordinary PCs. Almost none of it was ever written down, so I finally wrote it down myself. Back then real-time immersive 3D lived in exactly one place: industrial visualization labs running SGI hardware. An Onyx2 was a refrigerator-sized machine that cost millions. If you wanted a room people could walk into a dataset inside, that was the ticket price. I was at EON Reality in Irvine, California. In 1999 we built the Concave. Three Windows NT PCs. Three first-generation NVIDIA Quadro cards, about $1,200 apiece. A custom cable physically linking those three cards so all three machines swapped frame buffers within microseconds of each other. It had an RJ45 end on it, the same connector as a network cable, and it was not networking. Three Barco projectors onto a 120-degree curved screen with a 10-degree overlap, all the geometric warping handled inside the projector hardware so the PCs never spent a cycle on it. CrystalEyes shutter glasses genlocked across the whole rig. There was no API for any of this. No documentation, no vendor certification program. I cold called NVIDIA and asked for the sync code. It came to me from the rep who got it from the developer team, and as far as NVIDIA knew nobody else had done it. I had done the same thing with ATI four months earlier. The whole room, screen and projectors and software and glasses, came to about $750,000. That is roughly what one well-equipped Onyx2 cost on its own. That number is the part people get wrong. The Concave was not cheap. It cost about what the machine it replaced cost. What changed is that it was built out of parts anyone could order, and within a few years the UNIX graphics supercomputer was a memory. We ran a scan of the tomb of Tutankhamun on it. Curved screen, no bezel, no edge, nothing telling your brain it was looking at a screen. We pushed the stereo into negative parallax and the golden death mask came off the wall and hung in the middle of the room. Grown adults reached out to touch it. Every single time. I have spent the twenty-seven years since in IT. The thing that stuck was not the technology. It was watching a company with genuinely excellent technology die anyway, and watching an entire product category get replaced by commodity parts nobody in the room took seriously yet. Almost none of this is in the record. Prosolvia died before the internet learned to remember things, and demo rooms do not publish proceedings. If you worked at EON Reality, Prosolvia, SGI or StereoGraphics in that era, I would like to hear from you, and I would like to be corrected where I have it wrong. I am trying to get this down properly while the people who were there still can. #VirtualReality #ComputerHistory #XR