Computer Graphics - Marc Hannah

MH: They were just starting to do graphical user interfaces.  And a lot of that was initially developed at Xerox PARC, Palo Alto, CA. They developed this machine called the Alto.  And it was like a personal work station with a graphical user interface [GUI].  

Marc Hannah (MH): In 1979…you had personal computers, like the IBM PC.  So you could do basic stuff on a personal computer. Character displays, and so no fancy graphics or anything. ..keyboards with printers, and so your interaction was either with, you know, a paper tape, a tape with holes punched in it to represent the program and data or these decks of cards, you know, with holes to represent the data that you would…feed into the computer.  And it would print stuff on a sheet of paper.  Or you may have a terminal where you're actually interacting with it, but, you know, basically, you're typing on a keyboard.  It's printing on a piece of paper and the computer responds by printing stuff on a sheet of paper.  

MH: ...I needed to work on something different; wasn't sure exactly what it was.  I just sort of in passing mentioned that I was interested in computer graphics...there's a guy who just started this summer who is, you know, doing graphics stuff, Jim Clark. Would you like to meet him? I sat down in his office and he outlined some stuff that he was thinking about…one was the Geometry Engine, this parallel architecture for doing some of the computations involved in 3-D  graphics.  And another part of it was sort of taking this 3-D representation and rendering it into a screen, a two-dimensional screen image.  

MH: They were just starting to do graphical user interfaces.  And a lot of that was initially developed at Xerox PARC, Palo Alto, CA. They developed this machine called the Alto.  And it was like a personal work station with a graphical user interface [GUI].  

MH: They were just starting to do graphical user interfaces.  And a lot of that was initially developed at Xerox PARC, Palo Alto, CA. They developed this machine called the Alto.  And it was like a personal work station with a graphical user interface [GUI].  

MH: And then what you want to do…with 3-D graphics is to be able to display some particular view of that cube.  And so you're--you need to map those coordinates from what is…a [real] world space to this, from this three-dimensional world space to this two-dimensional…screen space. So it's actually a projection of, you know, a particular view of that cube onto this two-dimensional plane.  

 MH:…And there's just, you know, matrix math… [In mathematics, a matrix is a rectangular array or table of numbers, symbols, or expressions, arranged in rows and columns.[2][3] For example, the dimensions of the matrix below are 2 × 3 (read "two by three"), because there are two rows and three columns:]

It allows for a number or an expression to be transformed through a standards set of procedures based upon the components of the matrix.

MH: You have this [real] world space that sort of goes everywhere, and the screen is fixed in size and so you have to basically clip out things that are outside of the view of this sort of viewing pyramid that is projected onto the, onto this two-dimensional screen.  The Geometry Engine does that mathematics. At first it does matrix operations that sort of convert it into a, sort of a normalized space, and then you clip out parts that are outside of this viewing region, and then you sort of map that onto the screen coordinates.  And that was so it would come out with, two-dimensional screen coordinates.  So that was the function of the Geometry Engine…after you have the two-dimensional screen coordinates, then there was another process that would happen which would be to rasterize…if you had a square that was, you know, and you just have four points that define the corners of the square, but you want that to be filled in with the color blue, on the screen there is a memory where each location in memory represents one point on the screen.  And so, so you wanna fill in all the little points that are inside the interior of that square.  That process is called rasterization and so they're set apart where it takes the end points of geometric shapes and sort of fills in all the points inside…once Silicon Graphics started…I took over sort of responsibility for the Geometry Engine as well and did further evolution of that architecture.


MH: What was unique about the Geometry Engine [was the object] was…to bring the cost of high-performance computing down to much lower levels.  I mean for me personally, I had this interest in flight simulators and I wanted…this flight simulator thing which requires very high performance…[brought] down to a cost where I could afford to have one in, you know, as basically a game machine in my basement.  This was along the lines of what we were trying to do, although at this point it was still, you know, it wasn't a thousand dollar system or a hundred dollar system.  It was tens of thousands of dollars, but compared with hundreds of thousands of dollars that it might have cost you to do this before. 


MH: ..Jim [Clark] got more serious about starting a company, you know, based on this technology… November of '81…that was when Silicon Graphics was actually incorporated…I was focused on the technology, so that business plan and trying to raise money and so forth, I wasn't involved in…we got our first round of venture capital funding. I believe…$800,000 from Mayfield Fund. I was focused on the Geometry Engine and sort of redesigning that to make it much smaller…it worked…the prototype we developed…but it was too large to be sort of commercially feasible and not as fast as it could be.  So my job was to make it smaller and faster.


MH: At a time where I think a lot of computing, especially high-end computing was concentrated in big machines… the graphics was done as a terminal, but, so in this timeframe, a lot of people were just using character terminals.  But if you were doing this 3-D development work, you had these special graphics terminals that were able to do 3-D rendering very quickly.  But most of the computing was done on a separate machine… you would send the graphical data down to this, a graphics terminal to be rendered… It was just starting the evolution to sort of self-contained work stations for doing, you know, sort of higher performance computing [like the] Alto which had a nice bit mapped screen…The mouse was another technology that was developed… And so there was the beginning of this evolution from computers where all the computing was centralized and you had terminals to more personal computers and personal work stations and so forth, so combining graphics and computing in the same box.


JH: October of '83 I think was when we first introduced the terminal.  And that was followed up fairly quickly with graphics work stations and…the initial target, the initial target, primary target market was computer-aided design, so people designing mechanical things…very early on, one of the demo programs…was a flight simulator…It was later networked to be dog fight, but, and, you know, it turned out that that became a fairly good-sized market for us.  And also special effects, film special effects…it may have been "the" first system went to a company called Robert Abel and Associates who did film special effects.  And they, that became an area that sort of, Silicon Graphics became very well known for, and we dominated basically that field of 3-D special effects.

From 1995 to 2002 for eight consecutive years, all films nominated for an Academy Award for distinguished achievement in visual effects were created on Silicon Graphics computer systems.


MH: …we must have been at least 90 percent of the market.  And we definitely had the most powerful computers for that kind of thing…You want the sort of, there's an inner activity that's required and performances required as you're creating the figures and the motions.  And so, you know, that's a very big part of it, is the ability to quickly define an object and tell the computer how you want it to move and then see how it moves and make adjustments to that.  So that's one big part of it, is this inner activity.  And another big part is just the raw performance required to--once you have the basic shapes and movements defined, to render that into something that looks realistic…But you need incredible amounts of computing power to do that.  You know, they might spend, you know, minutes computing each frame and, of course, there's like, they're running at twenty-four frames a second, I guess, and, you know, there're a lot of seconds in an hour or in a two-hour movie.

Among the films SGI provided the hardware for the effects include, Jurassic Park" [1993], "Hunt for Red October" [1990], "Beauty and the Beast" [1991], "Aladdin" [1992], "The Abyss" [1989], "Terminator II [Judgement Day]" [1991], and the helmets on Monday night football that would clash together.

MH: But, you know, more and more… [special effects] capability was pulled into a standard package that, you know, anybody could buy, any creative person and so you could do all kinds of really impressive things without having to write up a custom software…it just became more and more widespread as the software became more available and the hardware became faster and cheaper.


MH: Yeah, I made a fair amount of money… Silicon Graphics was a very successful company.  And so, and plus, you know, I'd started off as a member of the technical staff, but sort of over time, you know, was promoted from member of the technical staff, principle scientist, chief scientist, vice president and chief scientist.  So, you know, so I continued to get stock options and so forth.  So, you know, in the end, I certainly can't complain, you know.  I made, you know, millions of dollars in--at Silicon Graphics, and I was doing, you know, work that I really enjoyed and doing stuff that I felt had a major impact on, you know, society as a whole…the movie special effects stuff was very fun and gratifying.  And it's always nice to be able to talk to anybody and relate to stuff that you're producing, but perhaps even more important was, you know, the way our computers were used in engineering and scientific fields.



MH: The CAD-CAM, computer-aided design, computer-aided manufacturing, that was the largest market for SGI products… companies like, you know… Ford, GM, Chrysler, those, major aerospace companies like…Boeing, McDonnell Douglas, lots of…government contractors, doing military work, both on the flight simulation and the aircraft design side of things, used a lot of our technologies; medical uses…part of drug development can be, you know, understanding a molecule and sort of the shape of the molecule and how it may or may not bind with things you're interested in…early on we were focused on graphics…I think there was a big major push on just the purely computing side, maybe with the first product introduction might have been in 1988 or something…half of SGI was focused on the combination of, 3-D graphics and computing.  But there was another half of SGI's business that was purely high-performance scientific computing.  And then, so sort of any scientific or engineering application where you needed super computers to do the work SGI played a role.


MH: from '82 through ‘95-‘96, SGI was were growing great guns, growing very rapidly, both in terms of, you know, revenue and the size of the company…[In 1995 Silicon Graphics acquired Cray Inc., The Supercomputer Company]…I think that caused some sort of, just tactical missteps in terms of selling.  We had a quarter [where SGI fell below their profit target] and the stock price declined significantly. 


MH: …I was always focused on the lower end of the product line.  And as we grew into higher performance things, you know, and split into divisions, I was still focused on the lower end…I felt that, you know, we had issues of, in terms of our product strategy in dealing with sort of the inevitable growth and performance of personal computing…I was focused on trying to get our technology pushed into consumer applications and lower end applications, volume applications so that as, sort of the bottom up into graphics that, you know, we were participating in that.  But I think, you know, with the Cray acquisition, we just got sort of defocused… I was trying to push the company in one direction and couldn't really get any traction.  And so, late in '96 I decided to actually to leave Silicon Graphics... I actually left in '97 [with the title of vice president and chief scientist].


MH: It all revolves around sort of education…both in terms of, you know, personal success and contributing to society as a whole…the African American community tends to be in the lower economic classes, and, you know, education is sort of the way that you get out of those classes…nothing replaces a great teacher, but, you know, and there is a part of the problem that's motivating the students to do well…I've sort of always been interested more on the low end of things than the high end of things because I feel, you know, ultimately, the cheaper things are, the more people can afford to take advantage of them. 

For any questions, please contact David Trotman :


1519 O'farrell St. San Francisco, CA 94115