Hello everyone, it's Sven Andersson here. I started off as an ASIC designer and then moved over into the programmable world. Now I work as an ASIC/FPGA designer at the Realtime Embedded consulting company in Stockholm, Sweden.
I first got to know our editor in chief, Max, in 2007 when he wrote an article about my How to Design an FPGA From Scratch blog in EE Times. (Click here to see that article.)
The years went by (as they do), but we didn't get to meet in person until February 2012, when Max was the keynote speaker at FPGA Forum in Trondheim, Norway. A few months earlier, Max had suggested to me that I should present a paper at the conference, because this would give us a chance to finally meet and have a beer together after all our years of email conversation.
I started to think about a subject I could talk about at the conference, but I had a hard time coming up with something good. One thing I've found is that when you don't have something new to present, you start looking back. After more than 40 years in the electronic industry, I think I have a lot of interesting stories to tell, so that was what I decided to talk about. (If you are interested, you can click here to download a PDF containing the slides from my presentation.)
In the future, I intend to talk about my current experiments with a variety of soft-core processors implemented in FPGAs. I've been loading the Linux operating system into these cores and running benchmarks. But first I thought I would begin with a little "retrospective" and explain how I came to be where I am today.
Let's begin with a graphic illustrating my professional timeline as shown below:
On April 19, 1965, Electronics Magazine published a paper by Gordon Moore, co-founder of Intel, in which he made a prediction about the semiconductor industry that has become the stuff of legend. In a nutshell, Moore said: "The number of transistors incorporated in a chip will approximately double every 24 months." This has since come to be known as Moore's Law.
I've come to realize that the results of Moore's Law have been a major influence on my professional career. This is what has made my life as an electrical engineer an ever-changing story. Year after year, larger and more complex microprocessors or memory devices were created and presented to the market.
A master of science in electrical engineering
On September 8, 1969, I started to study Electrical Engineering at Kungliga Tekniska Högskolan (The Royal Institute of Technology) in Stockholm.
It took me four years to get my master of science degree, and during most of those years I never came closer to a computer than the punched cards we used to capture our programs:
These days it seems hard to believe just how "primitive" everything was. For example, the only calculating tools to which I had access were a slide rule and a mechanical calculator similar to the ones shown below:
It's also amazing to me to reflect back on how quickly things changed. In 1970, for example, the Swedish company Facit was the world's leader in mechanical calculators. Just a few years later it was all gone. The reason was the miniaturization of electronic components and the commoditization of electronic products. In 1972, HP introduced the world's first scientific pocket calculator -- the HP35. This cost $395, which was a completely unattainable sum for me at that time. In fact, in those days, that amount of money would have covered all of my expenses for an entire semester!
My master's thesis
Everything had really changed when it came time for me to write my Master's Thesis in 1974. This was due to the fact that a few years earlier in 1971, Intel had released a simple 4-bit CPU called the Intel 4004.
The 4004 employed approximately 2,300 transistors and was created using a 10μm PMOS process. The clock speed was only 740 KHz and the processor could execute 92,000 instructions-per-second. Delivered in a 16-pin ceramic dual in-line (DIL) package, this was the world's first single-chip microprocessor and it started a revolution.
For my master's thesis, I was lucky enough to have access to a 4004 and to build one of the first microprocessor development systems. This consisted of the following components:
- 4004 -- 4-bit CPU
- 4008 -- 8-bit address latch
- 4009 -- Program and I/O converter to standard memory
- 1702A – 256 x 8-bit UV-erasable PROM
- 2602 – 1024-bit Static RAM
In order to create my programs, I captured them in assembly language on a PDP 11/8 and then used a cross-assembler to generate the corresponding machine code to run on my 4004-based system. A picture of a PDP 11/8 is shown below:
Following assembly, the machine-code program was output to a punched tape, and then read into the development system using a teletype with a punched tape reader as illustrated below:
When I consider how far we've come with things like smartphones, digital cameras, and today's tablet computers, and with everything connected using wireless networks, it amazes me to think that I used to have to transport programs from one computer to another using punched paper tapes.
Looking back, it seems like those were the Dark Ages, but it really was not so long ago when you come to think about it.
But now I want to know what you think. Are you interested in my recollections, in which case I would be delighted to continue with my "retrospective," or would you rather I jump forward to the present day and start talking about my experiments with soft-core processors in FPGAs?