1. Introduction

Recently, I was perusing through the equipment bin here at home and I came across an old AGP video card that had been sitting in it's bag for awhile now collecting dust. Curious to see which decelerator I had in there, I quickly realized I was staring at one of my old Intel i740 AGP video cards. In fact, it was a pre production board. Seeing this board brought back a ton of memories about my stay at Intel dating back to 1997 as well musings about the state of the graphics industry. It's incredible to think how much has changed since I was a 22 yr old Comp. Sci. graduate starting in a huge 60,000+ person company (at the time) and yet in many ways.. how so much has not changed. I thought it would be neat to write down all the various thoughts that I had as well as provide some historical insight into the i740 and what a simple graphics chipset did to the industry at the time.

2. A Young in…

I graduated from graduate school back in December of 1996 and joined Intel Corporation in March of 1997. At the time, I joined a relatively new group within Intel that was focused on bringing an AGP product to the market place to compete against the likes of 3DFX and ATI. Nvidia was still a relative newcomer and the real beast was of course 3DFX. The Intel i740 was codenamed "Auburn". Many thought that companies like 3DFX could not possibly compete against a company that had it's own fabs and a boat load of money. Ahh how wrong they were.

My own work centered around validating the OpenGL drivers being developed for the Intel i740 chipset. What the heck did I know about validation? Nothing really except that I knew it was not a field in computer science one wanted to stick around in for too long. However validating the OpenGL drivers was actually a very cool experience since I was quickly exposed to all sorts of new hardware that Intel was developing at the time as well as work from game companies developing OpenGL solutions. I of course also had time to develop my own OpenGL based test suites. All of this brings to bear what the industry was like during these days.

3. The Industry at the time

Do you remember 1997/1998? The big thing in those days was not so much who had the coolest 3D card. In many ways, 3DFX ruled the nest and the real battle lay in the graphics API standards. Were you going to develop for OpenGL or the fledgling Direct3D (or DirectX in general), or Glide? There were a ton of conspiracy theories floating all around with how Microsoft was sabotaging OpenGL by not providing an easy driver model for OpenGL as it did for DirectX drivers. Basically there were two ways to get hardware acceleration for OpenGL. Either you wrote the MiniClient Driver (MCD) or you wrote the Installable Client Driver or ICD. Essentially the easiest way to bring OpenGL to the market was to implement the MCD which allowed for a partial implementation of OpenGL while the rest was punted back to software implementations. The harder method was the ICD method where you essentially would have to implement a compliant driver following the Microsoft ICD interface which was a royal pain in the @#!#$@!$#.

So in essence, to get OpenGL to market was a real fuss and if it wasn't for many games refusing to necessarily cave into Microsoft's DirectX camp, I'm sure you'd see OpenGL less in gaming than you do now. However, in any event, the industry focused its efforts into this whole API battle. Yes there was also Glide but in reality it was never a major player as DirectX or OpenGL because of the proprietary nature of the API.

On the hardware front, the AGP graphics market was very new at the time with many wondering why AGP was even necessary. Memory was getting cheap and having tons of onboard VRAM was not going to be a big deal. At the time, one of the major advantages of AGP was along the lines that you could use system memory to store texture data etc. This of course isn't all that important these days but was one of AGP's inspirations was in this department. Another major important benefit of AGP was to relieve performance bottlenecks in the PCI bus attributed to graphics. As time has passed, this has been the single major benefit of AGP to the world of consumer graphics.

Intel had previously tried to enter the graphics markets a few times before and failed miserably. This 3rd time would be the charm - or so Intel had hoped. Why? Because this time, Intel teamed up with two other companies to help build this graphics card code named "Auburn". These two companies were Real3D and Chips and Technologies.

So who are the major competitors at this point? The usual suspects at the time were involved - 3DFX, S3, ATI, NVIDIA, and Matrox. There was NEC with their PowerVR but it really didn't make much of an impact. The real battle for Intel was going to be 3DFX and possibly NVIDIA. S3 was falling away at this point and ATI seemed less interested in 3D than it did in making sure it ruled the OEM market. So with the might of Intel and it's dominance in microprocessors and chipsets - why not rule the graphics chipsets market too? No reason not to think so right?

[Home] [Next Page]