The above picture shows the old Pixel office. It looks cleaner today that I remember it – somehow less run-down. Maybe it’s the lack of the big white lettering spelling PIXEL SEMICONDUCTOR.

I can remember the exact day that I started working at Pixel Semiconductor because it was the day that the branch Davidians set fire to their compound in Waco, TX (killing everybody).

It is a big shock going from a big company to a very small company in many ways. I took the job because it would give me more Windows programming experience as well as low level hardware experience. At the time, Windows VxDs (virtual device drivers) were all the rage and I’d get to work on them. I already had a little experience with VxDs from my work on the Windows Sound System – the engineer that wrote that VxD literally wrote the book on VxDs (we later worked together).

But the real challenges of working at a small startup were:

• Helping set direction when your boss is a “visionary” who has about 20 brain farts a day – all of which he thinks are brilliant ideas (hint: they were not).
• Working without a net. At a small company, there’s no team of people to steer you on the right direction or to compensate for your mistakes.
• Doing whatever it takes to keep the lights on.

Pixel made some video processors that were kind of cool, but also kind of worthless. These were still the early days of computing and most PCs didn’t have enough power to even display a JPG, let alone play a video. These video processors were on their own standalone board, but could be connected to most graphics cards via an external cable. You could connect a video stream to the card and it could basically display live TV in a window on your computer screen. We were hoping to capitalize on:

• Acceleration playing media within Microsoft’s Video for Windows.
• People that wanted to watch TV on their computer (unsurprisingly, not many people would pay for this).
• We wanted to sell chips to people making video editors (we wanted to compete with the Video Toaster).

The first challenge was figuring out how to tie into Windows to make any of this work. Video for Windows had some documentation, but it just was not setup for what we needed. I bought a copy of Andrew Shulman’s book Undocumented Windows and used that as a guide. Andrew did not work for Microsoft (at the time), but using various tools was able to reverse engineer exactly how Windows worked. His book provided some of those tools, so I was able to reverse engineered Video for Windows.

My job at Pixel required a lot of travel – most of it to provide tech support for sales. It was hard travel because we never had budget for hotels – we always had to fly in early and fly out late. Also, we always travelled with large PCs that normally didn’t travel well. It normally took an hour or two to get the system working once we reached our destination. It was often very frustrating.

We did have two very promising business prospects:

• Cirrus Logic at the time was a very large chip maker and owned about 60% of the graphics market. They made an investment in Pixel and were considering buying us outright.
• Compaq was interested in buying one of our chips that would allow them to play live video on their PCs. The problem was that the chip wasn’t ready and didn’t really work. They kept setting deadlines and we kept missing them.

Things came to a head around the Christmas of 1993. Compaq said that they’d back out of the deal unless we showed the chip working within one week. Cirrus Logic said that they would drop us if we lost the Compaq deal. So we were golden if we got the chip working within a week, otherwise, we were completely out of business.

Everybody came to work on Christmas day. A hardware engineer (Bruce Doyle) was going to California to perform laser surgery on the chip to make it work, so we spent Christmas day brainstorming how to maximize the chances. I wrote some software that connected to some hardware that would turn off the bus in certain ways to eliminate noise. The next day I drove to St. Louis to meet my fiancé’s family, and they were very nervous when I told them I wasn’t sure whether I had a job or not (and very confused that I didn’t seem concerned). I had no idea how things went until I got back to Plano a week later.

A Christmas miracle had occurred! The chip worked (thanks Bruce!), Compaq paid us lots of money, and we were fully acquired by Cirrus Logic. I was back to working for a larger company again (but still in an ugly business park).