Is there a market for on-chip instrumentation?
Verification and validation of completed chips used to be quite an easy task, at least compared to today. The processor, memories and peripherals were connected at the board level and bench test equipment, such as digital logic analyzers, oscilloscopes and in-circuit emulators could be connected to the board to enable you to see what was going on in the design. That was the environment I grew up with and given that I worked in the aerospace sector we also had to think about removing single points of failure by adding redundancy and including self-test circuitry so help software perform a complete test and diagnosis of the board. In fact the very first piece of software I wrote was test software for a flight controller that was being developed for the Airbus A310. This gave me a very good insight into the design and how it worked.
Fast forward a few years, OK a few decades, and all of that has now been integrated onto a single chip. The bus and everything else is deeply embedded into the silicon. JTAG and other similar standards were developed that allow the notion of the in-circuit emulator to be replicated through an interface to the external world. It is a frugal interface using only 4 pins in most cases and even when multiple processors exist on the chip, the same interface can daisy chain them together. But so far, there is no such standard for on-chip instrumentation although IEEE P1687 Internal JTAG hopes to remedy that.
But what about the rest of the system? There have been a number of other companies that created IP libraries for on-chip instrumentation. This includes data gathering circuitry for both analog and digital, compression logic, breakpointing and filtering and an interface to either get this data out in real time or store it in on-chip memories. For some companies this on-chip instrumentation logic could be reconfigured to become statistical and performance monitoring circuitry for production chips and one company even allowed the logic to be used to fix some kinds of problems in a chip. And yet each of those companies struggled and eventually got acquired, the technology seeming to fade over time. New companies have come along trying to fill this market niche. Just over a year ago, Tektronix who is one of the largest makers of bench test equipment bought Veridae, a company making on-chip instrumentation. This clearly says that Tektronix is somewhat concerned that this is cannibalizing their market and the best defense is to become the leader in the very technology that seeks to make you redundant.
Neal Stollon, one of the pioneers in this field, has even penned a book about the subject titled “On-Chip Instrumentation: Design and Debug for Systems on Chip” indicating that there is significant depth to this subject and perhaps I will try and do a book review or excerpt of this in the future.
In my mind, this creates some interesting questions about the market. First, what percentage of designs actually includes on-chip instrumentation, excluding JTAG? Given that no company seems to claim to be the market leader in supplying IP for on-chip instrumentation, then do companies develop this logic themselves? Given that this logic would seem to hold little commercial value in the final product, what percentage of chip area can be dedicated to a function such as this? Is there value in having 3rd party companies provide this or is the integration of it so difficult that it provides no savings of time and effort in the long run?
Brian Bailey – keeping you covered
If you liked this feature, and would like to see a weekly or bi-weekly collection of related features delivered directly to your inbox, sign up for the IC Design newsletter.