There is a new paradigm for silicon vendors offering processor intellectual property (IP). Add flexibility to the well- known mantra of "faster, cheaper and less power." To succeed in the competitive world of processor-based system-on-a-chip (SoC) design, chip manufacturers have to change their focus from developing "one chip for one purpose," a traditional ASIC, to offering a chip that is application-adaptable. This means the processor IP the chip uses has to be extendable and configurable to support the application flexibility of the end system.
Single-function ASIC chips are too expensive and take too long to develop. To justify the cost in time and money it takes to develop an SoC, the chip vendor is banking on two things. First, the specifications to which the chip was originally designed will not change during the development cycle and, second, that the supplier will sell a lot of chips to justify the development NRE. This is a gamble that few chip vendors can afford to take.
If end-equipment manufacturers understood the risks and inefficiencies involved, they would never design with generic, off-the-shelf CPU IP. A much better idea is chips that support a variety of applications through the use of extendable and configurable processor IP. Such IP produces a device that can be configured for a variety of applications within a specific technological domain, such as graphics processing or wireless communications, or even between domains. This concept is key to developing chips that meet the consumer-driven requirements of today's products at reasonable price points.
Configurable and extendable IP gives the chip developer several advantages. Since the end-system vendor can use the same chip for several systems, the high cost of chip development is amortized among multiple end products and their cumulative sales volumes. These end systems address application convergence — products that can efficiently deal with several technologies, such as voice, video and data, concurrently and with a single processing engine. Examples of where technology convergence is becoming increasingly important include wireless handsets that transition between Wi-Fi hotspots and various cellular networks, and automobiles where single-processor-based chips control comfort, communications, safety and engine-efficiency subsystems. Finally, the use of configurable and extendable CPUs assures that the chips in which they are embedded can deal with the variety of emerging and changing technology standards that are pervasive throughout the communications, data processing and networking industries.
Processing IP that is software extendable and configurable is the inevitable next step in embedded-microprocessor technology. Chip vendors are realizing that times have changed — product configurability is not a luxury, but a requirement for those that want to succeed in today's highly competitive market. Toward this end, processor-IP vendors have to provide processing cores with the function flexibility necessary to enable chip and system vendors to design the right products. Extendable and configurable processors are a way of protecting and enhancing the value of critical IP and of ensuring successful and ultimately profitable SoC and software product development.
Carl Schlachte, Chief Executive Officer, ARC International, San Jose, Calif.