We've all faced the challenge of solving a new complex computational problem with a need for more processing power than any one device could handle. Multicore devices and FPGAs can help. But for a tightly coupled, fault-tolerant, and scalable solution, a multiple processor cache coherent non-uniform memory access (ccNUMA) interconnect is ideal.
The RapidIO interconnect fabric technology is unique in its design as a low-overhead, small-footprint, edge-of-the-processor I/O. It's the number one choice of designers for DSP and CPU farms. Also, it has the largest ecosystem of semiconductors, IP, boards, and development systems in the market. And let's not forget software. RapidIO supports a shared-memory mapped architecture and message passing with all of the protocol for traffic management in the hardware.
Ready-to-use IP cores have been developed, and many vendors offer proven hardware. This IP can target an ASIC or any of the leading FPGA vendors. Just add DSP or CPU cores (as many as you like) and a RapidIO interface. A RapidIO Bus function model is available from RapidIO so you can perform pre-silicon verification.
DSPs and CPUs with native RapidIO interfaces provide a level of very tightly coupled devices that can only be created with the features of RapidIO. Many designs also have been completed with off-the-shelf processors interfaced to an FPGA or ASIC with a RapidIO interface to the backplane.
Designing a chip not your game? Build a board. With four switch vendors and two of the biggest CPU and DSP manufactures in production, designing a board couldn't be easier. The question becomes how many DSPs you can get on the board. Your software team will be happy to know there's no stack to manage.
Grab a bunch of boards and put them in a chassis. If your game is system integration, RapidIO has your back. VME, ATCA, and AMC— pick your industry-standard backplane or connector and shop the hundreds of ready-to-ship boards. Development platforms and RapidIO software for initialization and configuration are ready to go.
The RapidIO Trade Association was formed to develop a specification and promote the common business interest of its members. These members are only one of the keys to the tremendous success of this international industry standard. They're the providers to the networking, telecom, storage, signal-processor, and high-performance embedded applications industries.
Members with a common interest alone don't make a successful association. It also doesn't ensure that the other necessary elements are in place to first, create a viable and strong eco-system; second, support the creation of interoperable, scalable, and reliable I/O products; and third, develop international open industry standards for a high-performance, packet-switched, system-level interconnect with a viable, long-term roadmap.
RapidIO technology is based on a truly open, established, switched-fabric standard designed by embedded engineers for embedded engineers to power applications in military, industrial, wireless, triple-play, multimedia, and other communications markets. On top of that, it's designed to deliver the reliability, cost-effectiveness, performance, and scalability required by OEMs building next-generation equipment today.
The RapidIO Trade Association is governed by industry leaders AMCC, EMC, Ericsson, Freescale Semiconductor, Lucent Technologies, Mercury Computer Systems, PMC-Sierra, Texas Instruments, Tundra Semiconductor, and WindRiver.