Point Tools vs. Integrated Flows

Electronic News, April 30, 2001 by Gregory Hinckley

EVERYBODY LOVES THE idea of integrated flows, but no one likes the practical reality.

The reality is that customers have invested a vast amount of time and money in the tools they are already using. When it becomes necessary to make a change because a particular tool has reached a breaking point and can no longer perform the job, that change happens point tool by point tool. What historically does not happen is a wholesale swapout of all the tools in a flow.

There are five reasons for this, and none of them is minor. One reason is cost. It just doesn't make financial sense to replace everything when only one thing is broken. With a change of a single point tool, you can often solve your design problem without making a major investment.

The second reason comes under the heading of the old advice, "Don't borrow trouble"--only in this case it's more like "Don't buy trouble." When you plug in a replacement point tool, you now have to worry about interoperability issues on both sides of the place in the process where that tool operates. When you replace an entire tool set, even supposing all parts of the flow are truly integrated (a big if), you will still have interoperability issues--they're just in different places.

Then there's the people factor. When designers are under enormous pressure to design better products more quickly, they have better things to do with their time, in their view, than learn how to use new tools. It typically takes an engineer one to two years to learn to use new design tools effectively. Having to do that all the way along the design flow because you've replaced the entire tool set is enough to make any designer--and the project manager--shudder.

The fourth reason for not buying an integrated tool set is particularly compelling. EDA customer companies are in highly competitive businesses, where being first and most cost-effectively to market can be a matter of financial life or death.

To achieve their goals, customers need the best tools for the job at every step of the design process. There is no point at which they can really afford to settle for second best. Yet no EDA company can claim to make the best-in-class tool for each and every stage of the design. So any time you decide on an integrated tool set from a particular vendor, you are inevitably accepting into your flow some tools that are not best-in-class. Can you really afford to stake your company's next product--and possibly your company's future--on something that's not top of the line all the way? Equally compelling is the risk involved on another front. The vendors who are offering integrated flows are doing so with proprietary interfaces. That means a customer is banking 100 percent, do or die, on that vendor's stability and ability to keep that flow the best available--forever. At this stage, it's impossible to tell how many of the proposed integrated flows will turn out to be dead-end approaches. That's an awful lot of eg gs to place in one very expensive basket.

Even assuming all the other negatives listed above did not apply, integrated flows have no chance of working unless, and until, there are open interfaces and standards. Without that, the much-vaunted integrated flow may turn out to be something of a Trojan horse--it looks wonderful, but once you take it inside your walls, it can cause untold damage.

Customers want the best available tool for every aspect of their particular design needs, at every stage of their design process. They want it wherever they can find it, no matter which vendor has it. And when they get it, they want it to work with all the other best-in-class tools from multiple vendors that they already have.

That's what customers want. And that's what they deserve. EDA vendors can best serve customers' needs by providing best-in-class point tools and by ensuring that all of those tools work well with other vendors' tools.

Gregory Hinckley is president and COO of Mentor Graphics Corp.