Intel Unveils Advanced Network Processor Technology

SAN JOSE, CA -- Intel Corporation announced new technology that will enable the company to deliver the first fully programmable network processors capable of processing data at speeds exceeding 10 gigabits per second (Gbps). Intel's network processor and technology will be unveiled during the Microprocessor Forum this week. Intel's network processor family consists of advanced, programmable devices that are used in networking equipment to rapidly manage and direct data moving across the Internet and corporate networks. The new technology enables higher performance at lower costs by eliminating the need for specialized coprocessors that determine the flow of data packets into and out of a network processor. Intel's approach combines multiple processing engines with an Intel(R) XScale(TM) microarchitecture processor on a single chip, providing equipment designers with the flexibility to apply processing power to the tasks that are most important for the device they are building. "Makers of networking equipment need flexible solutions to accommodate a wide spectrum of features such as quality of service and virtual private networking," said Nick Finamore, general manager of Intel's Network Processor Business Unit. "Our approach allows equipment designers to meet performance and feature requirements without sacrificing their cost targets. By bringing Intel's design and silicon expertise to network processing, we're able to build the most powerful and flexible network processors on the market." Intel's new network processor technology combines multiple processing engines, sophisticated software pipelining techniques, and distributed memory caching mechanisms to ensure sufficient headroom to handle the most demanding packet or cell processing requirements. The 10 Gbps performance milestone is a significant improvement over Intel's current network processors, which are capable of processing 622 million bits of data per second. The new technology extends the multiprocessing architecture of Intel's IXP1200 Network Processor family and will be used in network processors available next year. The new processors will dramatically accelerate packet or cell processing through the use of software pipeline technology that enables the processing load to be distributed among multiple processing engines. In addition, high-speed communications pathways built into Intel's next generation of network processors will allow processing engines to share information about each packet to boost efficiency and performance. Local caching (storing information briefly before processing) of packet information further enhances performance. In addition to the advanced processing engines, Intel's next generation of network processors will also include a low-power, high-performance Intel(R) XScale(TM) technology-based processor on the same chip. The XScale-based processor is designed to handle more complex processing tasks on packets that fall outside of the normal parameters of the primary processing engines. The XScale microarchitecture provides high-performance packet processing while consuming very little power, enabling more dense implementations of networking equipment and reducing costs. The new network processors will be built on Intel's leading 0.13-micron manufacturing process, boosting silicon performance to 1.4 GHz, reducing power consumption and lowering manufacturing cost. Intel's new technology will be implemented in new families of network processors, offering developers a choice of processors for communications equipment used in a range of settings, from offices and data centers to the core of the network. To minimize development time and cost, the products will be supported by a common development environment and tool suite based upon those used with the IXP1200 Network Processor. For further information visit www.intel.com