The NVM Insider, Issue 2 - Page 3 - Outside Thoughts

Outside Thoughts:  NVM Opportunities

Bill Martin, Mentor Graphics

The high technology community has always been innovative. Regardless of the technical or business issues that exist, clever individuals create or find solutions to these issues. As we entered the new millennium, four apparently opposing issues have hit products at the same time: rapid increases in product development costs, increasing worldwide pressure for ‘green’ products, faster moving and larger consumer markets, and increasing integration capabilities and challenges. Let’s examine each briefly and examine how Non-Volatile Memory (NVM) can be used for each.

Rising development costs

Over the past decades, technology and product development costs continue to rise. To combat these large increases, many people have come up with novel technical and/or business solutions to help offset these costs. A couple of examples include:

• Many large IDMs in the past year decided to become fabless, joining alliances for process development and leveraging a larger pool of cash. This concept has been used in the past by ‘pairs’ of companies, when neither had the finances to do it alone. This current model is different because many more companies are joining a common alliance and all are based on common equipment.
• FPGA providers create silicon solutions that leverage expensive mask costs among a larger customer base rather than a single mask set per design. With advanced process nodes requiring million-dollar mask sets, customers can dramatically lower their development costs. This might require performance and functionality tradeoffs for a SoC solution.

NVM can be an integral component on platform enablement. NVM can allow either the product developer or, possibly, the end consumer to configure a product.

“Green” is spreading

The “green” movement has been around for a long time. What has changed is the momentum of this initiative. More people are looking at methods to save energy or use non-oil fuels to live. A few examples are:

• Green PCs and workstations already exist and new developments on all products continue to evolve.
• More initiatives on wind and solar technologies are being developed
• There is more emphasis on fuel-efficient cars today

Just as design-for-test (DFT) and design-for-manufacturing (DFM) were initially hard and more expensive to implement, in time design-for-green (DFG) will reduce the overall system operating expenses. An individual component might be slightly more expensive, but this should be covered by system-level reductions. An example would be Silicon on Insulator (SOI) ICs. A little more expensive but for the same performance, you have a 20-40% reduction in power consumed, which will reduce system-level requirements. New design methodologies and tools are helping development teams understand how to reduce power without giving up either functionality or performance. NVM can also be used to permanently or temporarily turn-off functionality that is not required, thereby saving power and extending battery life.

Rapidly changing markets

Products geared for the mass markets normally are supported by a large infrastructure investment. If your product is late or does not gain customer acceptance, you could lose your product line or your company. Smart product companies will develop products that can adjust to evolving customer needs. Some examples might be:

• Products that are software-configurable allow customers to modify applications based on need; for example, the ability to “disable” applications that are not required in order to increase battery life.
• Products that are hardware-programmable allow customers or product providers to alter what is available in a product. This may not be as easy or as reversible as software-configurability, however.
• Another option would be self-configurable products, products that understand their environment and perform operations on their own. Some cars automatically open or turn-on based upon their owner’s proximity to the car.

Product-programmability allows developers to reduce their risks with minimal costs or time required. Again, NVM enables products to be configured for changing markets.

Increasing integration capabilities

Process nodes for today and tomorrow provide more integration possibilities than our current imagination can use. Often ICs dimensions are derived by the number of I/O pads required rather than from the internal core (gates) required. Few products provide enough logic to create non-pad-limited ICs, resulting in areas that are devoid of gates. Silicon providers do not provide discounts on unused area so, in effect, this area provides ‘free’ gates to the customer. Rather than wasting this area, developers can use it for functional purposes, as follows:

• To enable platform design, allowing several derivative products to be created from the same silicon.
• Provide extra functionality allowing regional requirements to be added at no additional cost. An example could be multi-banded handsets.
• Add ‘spare’ gates, allowing minor errors to be quickly fixed with simple metal/via mask changes.

Each of the above ideas allow developers to minimize the time and money required to release their product to market while expanding their potential customer base by meeting smaller, regionally-based functionality. NVM can be added with no increase in die size (most designs are pad limited) and provide flexibility at no additional cost.

NVM’s role?

NVM has made significant progress over the past decade. Years ago, it was relegated to older technologies and was more expensive to use. The expense was driven by different process requirements, more area required, and complex programming requirements. These barriers have dramatically fallen. Remember, we have lots of clever innovators in high technology. Provide a problem, they will determine a solution!

Given the above challenges and how NVM can be used for little, if any, additional cost, why would someone NOT consider using it in their product?

Bill Martin has over 25 years of experience in consulting, product design and project management. He joined Mentor in 2000 and has held senior roles in Mentor Consulting and IP. In addition to Mentor, Bill previously worked for Synopsys, VLSI Technology and Mostek. He holds an MBA from the University of Texas at Dallas, and a BS in Computer Engineering from the University of Illinois, Urbana. Bill has also been granted 5 patents.