Switching from Pixel to Cirrus Logic was not a drastic change, but it was impactful in that we moved from creating separate video processors to integrating our video technologies into existing Cirrus Logic graphics chips. As Cirrus Logic had 60% of the graphics market (at the time), it also allowed me to engage more formally with Microsoft. And we did eventually move into a much nicer office.
Our new office building
Both technologies focused on improving the Digital to Analog Converter (DAC) in the graphics hardware. The DAC is responsible for reading the image in the frame buffer many times each second and converting this image to an analog signal that is sent to the monitor. The improvements focused on enhancing the DAC so it can read multiple steams and mix them in real time.
One of the scenarios we called “hardware overlay.” If you are running an 8 bit (256 color) desktop, hardware overlays allowed the video playback software to write a 16 bit video image to a different part of the frame buffer, and then the hardware can blend it into a single signal sent to the monitor without losing any of the video fidelity.
The other scenario that we focused on was called “video port,” where a separate peripheral can stream a video signal straight into the DAC via an external connector (enabling full frame rate live TV).
Of the two scenarios, hardware overlays are still used today, albeit in a limited number of scenarios. For example, XBOX One relies heavily on this feature.
But none of this could work unless Windows allowed the right hooks for us to make it work. The first real attempt at this was a joint venture between Microsoft and Intel called the Display Control interface (DCI). I developed a relationship with the Intel people working on it and we were able to make it all work. I then became responsible for making all Cirrus Logic chips support DCI.
At one point I was made a manager over a team of developers and I really hated it (as did my employees).
DCI ws not a huge success and the problem it solved (accelerating video playback), was important but it wasn’t the most important issue facing Microsoft. The much bigger issue they were facing was that a very high percentage of DOS video games were being returned due to hardware incompatibilities, and Windows didn’t have an infrastructure to support gaming.
Three guys at Microsoft (Eric Engstrom, Craig Eisler, and Alex St, John) came up with the idea of DirectX to solve this. Together they were known as “the beastie boys.” Eric was the Program Manager (PM), Craig was the developer, and Alex was an evangelist. I knew Eric and Craig really well – Alex much less so.
Alex (left), Craig (back) and Eric (right)
The first DirectX component was called DirectDraw, which was designed to give games direct access to features supported by most graphics cards. While the primary focus is gaming, it also is used to accelerate video playback. I attended the first DirectDraw kick off meeting at Microsoft in December of 1994 and was able to influence its feature set.
In early 1995, Microsoft was ready for the graphs vendors to start creating drivers for the DirectDraw interface, so they invited the major graphics vendors to come to Microsoft for two weeks to get started. Each graphics vendor got their own office in Building 20 (at the time called the “porting lab”), while Eric and Craig went round all day answering questions. To me, everything seemed rather straight forward so I had no questions, but I could hear them up and down the hall having the same discussions for days on end. Towards the end of the first week, Eric poked his head in our office and asked if we had gotten anywhere at all (he assumed we hadn’t). I showed him a demo of our driver working and he was so shocked that he ran and got Craig to show him. From that time on, they really liked me.
Eric and Craig wanted me to spend the entirety of the Summer of 1996 on campus working on the Cirrus Logic drivers for DirectDraw. I worked out a deal with Cirrus where they’d fly me to Seattle on Sunday afternoon and then would fly me home to Plano on Friday afternoon. This went on for about three months, only this time I wasn’t hidden in the porting lab – I had a real office in Building 4 where Eric and Craig worked. Cirrus Logic was the only IHV that had an office on campus at that time.
Craig was the main developer, and Eric was the Program Manager (PM). Craig had a standing rule that he would answer any dumb question that I had if I asked it after 1:30am, and there were many nights where I took him up on this. I’m a workaholic, but these guys were insane!
Cirrus Logic was a very political place (far too political for it’s size). Their headquarters were in Fremont, CA, so I lost a lot of political capital by living in Texas. Many of the driver developers were upset that I was getting to do all of the fun work. Cirrus Logic decided to solve this problem by creating a dedicated team of people whose job it was to interact with Microsoft, and this team would be based in Bellevue, WA (just 2 miles from the Microsoft campus). They asked me to be on this team and I accepted.