There are two ways to look at the overall impact of FPGAs (large scale programmable logic technology) on the electronics industry. The first is to consider them as a cool new way of implementing digital logic. This is the conventional model which sees FPGAs as a better, cheaper, faster and field-configurable way of implementing digital electronics, and is part of the ongoing trend from discrete low-level chips, through PALs and onto today's massive FPGAs. This view is simple and comfortable - doing the same thing in a more efficient way. It is not really disruptive, it just gives us access to more power at lower cost. This approach is all about extracting the nth degree of optimization from FPGAs.
If, however, we look at the position of FPGAs at the bigger picture level, what we see is a technology that is part of a much larger trend in the development of electronic products - the move from 'hard' to 'soft' design. This is a more 'scary' view of the world because it threatens the way we view the whole process of designing electronics. It is disruptive because it forces us to reconsider why we do things a certain way.
The last "Big Thing"
The real clues as to what is really going on with FPGAs can be better understood if we first take a trip down memory lane. Let's go back to the early days of the microprocessor - perhaps the last real 'paradigm shift' to hit the electronics industry.
The first real microprocessors were developed by Intel on contract as 'computers' for a desktop calculator product. Although their development was not really part of any grand strategy (Intel was really a memory-chip company at the time), it wasn't long before their potential was spotted and the personal computer revolution ensued.
The widespread use of microprocessors as 'embedded controllers', directly replacing 'hard' electronics, only came after their decreasing cost made them competitive with the old 'hard-wired' solution. Once this point was hit, the use of microprocessors in electronic systems exploded! Embedded software development, as a way of describing the functionality of electronics systems, became a respectable, specialist part of the overall design process.
In general though, the important thing to look at here is not the microprocessor device itself, but rather the change in design paradigm that it enabled - the move of design content from the 'hard' to the 'soft' domain.
Why is the move from 'hard to soft' so important?
Once the design moves into the 'soft' domain, it brings some very strategic benefits. The design can be split into two major parts - the design of the hardware platform (the 'dumb' part) and the development of the actual device behavior (the 'intelligence'). This platform-based approach allows the same physical hardware to be manufactured with much lower risk because we can modify the behavior of the application after constructing the hardware. Other significant benefits include:
- Being able to develop a large part of behavior before we have the final hardware
- Once we have the 'real' hardware we can continue the development and debugging process on this platform without needing to re-manufacture as we make changes
- We can upgrade the device in the field by uploading new software for it
- Our inventory needs and device availability risks go down because a microprocessor can do many different things
- And, especially in recent times as embedded microprocessors have become so powerful and cheap, we see an explosion in the 'smarts' of devices - what we do with today's devices is no longer even possible with conventional hard-wired electronics.
Starting out life as a tool to increase the efficiency and scalability of designing electronics, the microprocessor ultimately rewrote the rules on what was demanded of an electronic device. It is no longer good enough to be cheap, reliable and efficient. Today's electronic devices need to be 'smart'. 'Dumb' hard-wired electronic devices need no longer apply!
So where do FPGAs fit in?
When considering the history and potential of FPGAs, the analogy to microprocessors is striking. With FPGAs the move from 'hard' to 'soft' design that started with the microprocessor revolution can now advance to the next level.
FPGAs started out life in the mid 1980's as a replacement for discrete logic, reducing chip count and improving our power to cost ratio in digital hardware design. Now, thanks to the inevitable march of Moore's law, these devices have the potential to once again rewrite the rule book on electronics design.
Microprocessors allowed device behavior and algorithms to be described in a soft way and then 'compiled' into the hardware. FPGAs now allow us to take all of the digital hardware and describe it in a soft way (including the actual microprocessors) and then compile this into the design.
Now that these devices are passing a crucial threshold in terms of offering a sufficiently large-scale platform at a low enough cost, it doesn't take an oracle to see where this is all going.
Large parts of the system design hardware will inevitably flow onto the 'soft' FPGA domain and join the processor software as part of the 'embedded intelligence' of the product, rather than being part of the hard-wired platform.
PCBs will be the physical platform and provide the connection between the 'real' design platform -- the FPGA -- and the outside world via analog interfaces. The current movement of so much of the 'analog processing' world to the digital domain will simply enhance this effect.
Arming ourselves for battle
The fact that such a large part of the 'real' design can now be described in a soft way and then compiled into the physical hardware forces us to review our whole approach to the design process. The traditional 'divide and conquer' tussle between hardware and software design is going to be completely inadequate in meeting the challenge of harnessing the power of the emerging paradigm.
The designer can now make choices about how to split the system between software and 'soft' hardware implementation after the physical device is actually manufactured - even after the device is in the field. To realize the true benefits that moving the entire system into the soft domain offers, design tools must support the convergence of software and hardware design within the transmutable environment of the FPGA. They must therefore support a much more holistic approach to the entire design process. Without this the benefits of working fully within a 'soft' environment cannot be realized.
Even to allow us to make efficient use large-scale FPGAs in the 'old' way as implementations of powerful but low-cost digital electronics, the PCB and FPGA design must move to a much higher level of integration/communication. This is something that is happening across the design industry today.
But in order to fully harness the power of this emerging 'soft' design paradigm, design tools must go a lot further.
In particular, the FPGA-hardware and embedded software domains must move towards becoming a 'singular', unified design process. When this happens brace yourselves for a state change in the way we do electronics design and an explosion in innovative applications to rival that heralded by the introduction of the microprocessor itself.
By Nick Martin, Joint CEO and Founder, Altium, Ltd.
Martin founded Altium Limited (previously known as Protel International) in 1985 with the single vision of breaking down the barriers to innovation by providing every engineer with easy access to state-of-the-art design tools.
Go to the Altium, Ltd. website to learn more.