Vegard Wollan talks about the history of AVR

We were lucky enough to drag Vegard Wollan into the Atmel Studio while he was visiting headquarters from Norway a few weeks ago. For those of you who may not know, Vegard is commonly believed to be the “V” in AVR. After having spent hours in the Studio, we decided to break the film into separate segments — this one is about the history of the AVR. And, what better day to recount the earliest days of the revolutionary microcontroller than Throwback Thursday?

You young whippersnappers might not realize it, but Atmel started life as a memory company. We made EEPROM memory, and soon branched into Flash memory. Vegard and his pal Alf Bogen (the “A” in AVR) met on computer forums while students at Norges teknisk-naturvitenskapelige universitet, also called NTNU, or the Norwegian University of Science and Technology. They saw the need for a microcontroller based on Flash memory, which you could reprogram as many times as needed, even when it was in-circuit.

AVR inventor Vegard Wollan holds his Master’s thesis from college that began the AVR MCU architecture.

This was at the time where development engineers had to buy expensive ceramic package microcontrollers with a quartz window. You would program the part with your Data IO programmer, and when you had to make any change, you took the part out of its socket and put it in an ultraviolet eraser for 20 minutes. Savagery.

So Vegard and Alf pitched their MCU design to Atmel, who immediately saw the great potential in it. Atmel had already started to make 8051-style MCUs in the Colorado fab facility, which we had just purchased from Honeywell. One great thing about those parts was most instructions executed in one cycle instead of the 4 to 12 that the original 8051 needed.

So when Vegard explained the AVR was a RISC (reduced instruction set) machine designed to execute instruction in one cycle, it was music to Atmel’s ears. You could say that AVR was RISC before RISC was cool. And, it explains why AVR has passionate followers just like ARM-core RISC chips also have passionate followers… I guess that explains why Atmel makes 8- and 32-bit AVR chips, as well as a whole line of Atmel | SMART ARM-based chips.

The other cool thing that this video touches upon is how AVR chips were designed to run C programs well. You might remember the early days, when Intel would make whatever hardware people thought was cool and Microsoft would program in ways that software people thought was cool. Things didn’t always line up. I remember how you had to do a software reboot to take a 80286 out of protected mode, or that “thunking” in Win 95 to switch between 16- and 32-bit.

Vegard is a Renaissance Man that understood the hardware implications of good software design. So he made sure that C code would compile superbly into AVR assembly language. My buddy Wayne Yamaguchi routinely writes C programs for AVR that compile down into a few hundred bytes. AVR is one of those magnificent examples of computer science, not computer “slap this together and push this out so we can sell it”.

Like all Norwegians, Vegard is incredibly modest about his contribution. He credits his co-founder, and the team at Atmel that developed the AVR. But you just have to look at the billions of AVR chips that Atmel has sold to see what a remarkable thing Vegard created. Check out the video and stay tuned for the next installment.