Dean Brown: Building Legends

Dean Brown got his first Amiga 1000 in early February of 1986, coming to the platform after several years running his own consulting and programming business developing for the Tandy Color Computer, Model III, and a MS-DOS clone from Sanyo. Frustrations with ‘vapor product’ led directly to what would become the legendary DKB Insider, a 1-Mb memory expansion card for the original A1000 computer. The Insider was one of the first peripherals available for this new platform, and the response proved prophetic.

“In early February of 1987, we built 250 boards expecting this to be sufficient for four months of sales, but the entire production run was sold out in two weeks,” says Brown, and the response was world-wide. “I was getting calls from people all over the world-France, Japan, South Africa-which to me was just flabbergasting.” Selling in excess of 6000 Insider boards in 1987 defined the true beginnings of DKB Software.

The Insider

That success continued with other products, including the MegaChip and MultiStart that were dear to many Amiga owners’ hearts. “I was at one point selling 1700 MultiStarts a month,” he says. “They went in, did the job, and people forgot they were there, for the most part.” In many ways, this is the definition of reliability, a topic Brown is passionate about. “In designing a product, you do a timing analysis to find out whether or not things are going to respond within the correct amount of time,” he explains.

The MultiStart

"What I noticed was there were a considerable number of products out in the marketplace using typical timings of the the componenets, rather than worst case. If you go to worst-case timings and make certain your product works under worst-case circumstances, it will be a more reliable product." As personal computers see huge varieties of less-than-ideal temperatures, electrical service, and operating environments, that reliability becomes essential.

The DKB 1202

Superior reliability is great, but people still have to buy your products. When the Amiga market dried up during the ‘dark time’ after Commodore and before Bill McEwen, Brown was forced to seek alternatives. “I can't say I ever abandoned the Amiga market, but I did de-emphasize it,” he says. “In '97 I had to close the doors of DKB as it was no longer supporting my family. For almost two years I supported the market as best I could through arrangements with Joe Rothman of Mr. Hardware Computers.” But when Rothman died, that essentially ended.

Brown’s passion for the platform remained, and again found purchase in the formation of the Industry Council on Open Amiga, or ICOA. “The objective of the group was basically to put together a set of standards by which people could design the next generation Amiga. ‘This is the way we’re going to do things; this is the standard we’re going to set,’ explains Brown. And everybody’s going to work toward that standard.” This focus shifted slightly when Gateway got involved. “The thought then was that we could provide our services as an interface with developers and provide market intelligence about what people were looking for.”

Things initially looked positive. In August 1997, the ICOA Steering Committee was flown out to Gateway to present their case. Because Gateway showed an interest, the ICOA backed off a little. “If we had a company that was going to be driving the machine forward, the ICOA’s basic focus would no longer be as critical,” says Brown. “Unfortunately, it came out that Gateway had no clue what they were doing themselves, nor any interest in furthering the Amiga marketplace, and the end result was the ICOA died a slow and painful death.”

Some positive did come out of that phase, at least for Brown. He met Fleecy Moss, first online, then in person at a dealer open house. “We hit it off right away,” he says. “We have many dreams in common.” Like space exploration and spicy food. And the possibilities of computers, “like how and why the machines of today are not doing the job they should be doing,” he says. “For example, user-interfaces are really targeted at tech-heads.” He goes on emphatically. “If you’re trying to break into a much larger consumer marketplace, you have to provide an interface that’s comfortable for people to use.”

The synergy Brown and Moss felt stuck with them. “Just prior to Amino Development purchasing the Amiga assets from Gateway,” Dean explains, “Fleecy and I began discussing future directions again. It really came as no surprise when I was asked to work with Amiga.” Brown first contracted with Amino/Amiga, helping to define and develop Amiga’s first products. Then came an offer to work full time. “As of June 5, I am now a full time employee,” he says. And his position will fittingly be as Director of Hardware.

“You become a champion by fighting one more round. When things are tough, you fight one more round.” - James J. Corbett

The job has several responsibilities. “First and foremost, my job is to create the test platforms that will be used by the internal software teams,” Brown explains. This will result (initially) in systems that look very similar to existing computers. “These same platforms, or modifications of them, will be made available to our hardware partners as reference designs. In the medium term, I expect to develop a new architecture that addresses many of the issues that plague current computer systems. Issues that in inflexible operating systems cannot be fixed,” says Brown.

“If you look at architectures today, you see a whole bunch of processors running at the gigahertz range. If you actually benchmark performance between a 500 MHz processor and a 1 GHz processor, you’ll find the actual increase in performance is very small,” says Brown. Small to where doubling the clock rate only results in maybe 20% faster performance. The reason is simple. “The gating factor is not the speed of the processor. It’s how fast can you move data in and off the system.” What’s worse, he continues, is the current trend to drive prices down by speeding up the central processor and having it do more work. It’s a centralization trend disturbingly analogous to mainframes versus personal computers.

“Distributed processing is where I want to go.” There are lots of problems, he says, “But quite honestly, Tao has a big chunk of the problem solved.” Tao is able to run software on multiple processors without changing the base code. “At load time, the Virtual Processor translator translates from the VP code to native code, which means that which translator it’s loaded with determines which processor it’s running on. That’s a very powerful thing,” Dean points out. “The end result is we can have multiple processors in a single system, and the code can run on whichever processor is best suited for handling it, all decided at load time.” Additionally, he explains, where a CPU-centric system has to actively pass data multiple times to execute one action, in Tao’s distributed model, the only concern is that the data transfer happened. This allows the use of multiple, smaller, cheaper processors that collectively deliver performance exceeding a ‘faster’ CPU.

Distributed processing is philosophically similar to the classic Amiga architecture with its custom chips, though there are fundamental differences. “The custom chips were very specialized, architecturally tied to the operating system,” Brown explains, “so any time you wanted to make a [hardware] change you had to completely rewrite the OS.” Combined with Commodore’s misguided encouragement of code that accessed the hardware directly, what resulted was essentially an upgrade deadlock.

“We’re not going to let that happen this time. Commodore's handling of the Classic Amiga was haunted by the attitude that all legacy applications had to run on multiple OS and hardware versions. This prevented the Classic Amiga from taking advantage of newer technology,” he adds. The abstraction layer methodology central to the new architecture will serve this ‘backward compatibility’ purpose in the new Amiga, an idea Brown is passionate about. “I promote that applications that attempt to go around these abstraction layers be deliberately broken!

“Shortly after we release our first real hardware that’s natively running the new operating system, there will be another machine that is hardware incompatible with it running the same OS.” Applications will be tested on both side by side. “Our precedent here is: If it works on both of these machines, we know it’s not banging the hardware,” and therefore a truly ‘OS-legal’ program, he says.

"Everything that can be invented has been invented." - Charles H. Duell, Commissioner, U.S. Office of Patents, 1899.

“We’re trying to create something that’s brand new,” he says. “There will be some missteps along the way, I’m sure of it. As far as I’m concerned, a truly progressive company is the one that’s able to look at it, determine what those missteps are, correct them and move on,” explains Dean. He sure sounds like his feet are firmly planted on the ground.

It’s somehow fitting that Dean Brown, long-time Amiga fan and developer, one of the first Amiga icons, is part of the new effort, poised to influence what very well could redefine computing, as we know it. Lofty aspirations, perhaps, yet we don’t usually accuse legends of striving to underachieve. “I 'came back' to the Amiga because of the opportunity to make it something that is far more than what is available today,” says Brown. “We have an opportunity to break free of the baggage that retards progress, and I for one am thrilled to be a part of that!”