For me, it struck home when even my mother, who lived her life in technical obscurity, worried about how the floating point bug might prevent her from using her computer. As the world scrambled to fix the bug, it brought to light the necessity of hardware manufacturers and tool vendors to provide better solutions that ensure that their new hardware worked correctly.

Enter the need for faster simulation, assertion-based verification, smarter testbenches, better verification methodology and a slew of supporting products and technology.

It was simple economics. Manufacturers found it crucial to expose and fix these bugs before they found their way into chips and millions of consumer products. That's because re-engineering and fixing all the broken hardware is expensive, time consuming and a public relations nightmare.

Fast forward 16 years and read today's headlines. Massive automotive recalls bringing an even greater public awareness and anxiety about software problems affecting mission-critical systems. Only this time it's not a program on your computer that can fail, but systems that ensure your family's safety!

Software? Mission Critical to the average person? How can this be? Software was something that didn't have to be correct the first time. You could always patch the software and get it out to consumers with a new download or service pack update.

Things have certainly changed, haven't they? In the case of cars, what once was in the domain of mechanical engineers to ensure the safety and integrity of critical systems has now transitioned into the domain of embedded processors and software. Do you think patching all the software in millions of cars worldwide will be easy?

Again, using my mother as a litmus test, she certainly understands the magnitude of the recent auto recalls and the danger of braking systems that could fail. Of course, this doesn't just apply only to the automotive industry. More and more of what the average person depends upon and relies on is increasingly controlled by software running on embedded processors. Will this recent technical pandemic cause another call to arms of systems manufacturers and tool vendors to provide better solutions? I certainly hope so! Re-engineering and fixing all the broken software is expensive, time consuming and think of ensuing public relations disaster. Sound familiar? Pile on top of that the danger of having these systems fail in the hands — or feet, in this case — of the consumers. Now the stakes are even higher as the cost of getting it wrong creates critical problems in the daily lives of consumers.

The question we should now ask ourselves: How can we ensure that the very software that surrounds us — and that we depend upon — works as intended? I would hate to be the poor guy who finds out that the software he depended upon didn't do the job when he needed it. Oh sure, they could roll out a fix quickly. Maybe it will save the next guy!

The semiconductor industry needs a much greater focus on making sure software and systems work correctly together upfront, ensuring that products operate as intended before getting into consumer's hands. The industry cannot afford to take the haphazard approaches with software as it has in the past.

Better system-level tools are needed to bring software and hardware designers together earlier. Better and more accurate modeling is needed to ensure software engineers can understand the affects that the hardware and software have upon each other. Software needs to be tested on platforms that provide 100 percent accuracy with respect to the final product. Today, the risk of failure is much too high for the semi-passive approach the industry has taken to make sure software works in embedded systems. Just reading the headlines attests to that.

This means that engineers need to have access to tools and models that ensure the software they are designing will run as intended when the products are in consumer's hands. Perhaps, this is the push that system-level and platform-based tool providers — or the EDA Consortium, the Global Semiconductor Alliance or the Semiconductor Industry Association — need to shed the appropriate light on these critical issues. Because it isn't a simple matter of dropping a few calls on your cell phone any more. The stakes are much, much higher than that.

It comes down to this: Do you want grandma's car to stop in the parking lot or 15 feet inside the Quickie Mart? To make sure grandma's car stops in the parking lot, let's get it right the first time!

About the author:
Andy Ladd is vice president Applications and Methodology at Carbon Design Systems (Acton, Mass.)