Embedded Boards feature shared memory architecture

SCRAMNet GT PMC, PCI, and VME boards deliver real-time control and data networking, and enable networked heterogeneous computers to function as single multi-processor system. They feature 2.5 Gbit/sec SFP transceiver with optional secondary SFP available to provide redundancy. Shared memory on each board starts at 128 Mb and supports data throughputs surpassing 200 Mbps. SCRAMNet GT boards support Programmed Input/Output for small data transfers.

********************

New 2.5Gb/s SCRAMNet GT boards deliver data 13 times faster and feature new low, retail price

Leesburg, VA January, 17 2005 - Curtiss-Wright Controls Embedded Computing (CWCEC) has announced a new family of embedded boards based on SCRAMNet GT, the new generation of its popular shared memory data communications architecture. The new SCRAMNet GT PMC, PCI and VME boards deliver real-time control and data networking 1200% faster than boards based on its predecessor, SCRAMNet+, and enable networked heterogeneous computers to function as a single multi-processor system.

"The launch of SCRAMNet GT highlights CWCEC's ongoing commitment to maintaining both technology and value leadership in the 'benign' environment segment of the embedded market as well as in the COTS sector," said Lynn Patterson, vice president and general manager of Modular Solutions, CWCEC. "Because SCRAMNet GT simultaneously handles network control functions and network data streams, designers of networks for simulation, virtual reality and other demanding real-time applications can create a single, high-speed, low-latency data link that eliminates the need for multiple networks in the same real-time system. This, along with significant performance increases and a retail selling price at half the previous generation, dramatically improves the cost/performance ratio."
Advertisement

Each SCRAMNet GT board features an industry standard 2.5 Gbit per second Small Form Factor Pluggable (SFP) transceiver and delivers data at throughput rates exceeding 200 Mbytes per second. A secondary on-board SFP is available as an option to provide redundancy. Networks of various sizes can be supported through the use of two types of SFP transceivers; short-wave/multi-mode or long-wave/single mode. The short-wave/multi-mode SFP supports connections up to 300 meters (using standard fiber), while the long-wave/single-mode SFP supports connections up to 30 kilometers (using long-link fiber). Both types of transceivers can reside on a single board. The boards are also designed to support newer types of SFP transceivers (copper, 1500 nm, etc.) as they become available.

"SCRAMNet GT is a major step forward in the continued enhancement of the popular and successful SCRAMNet architecture," added Ms. Patterson. "Deploying SCRAMNet GT on the industry's leading open-standard embedded platforms will simplify network and control system design for countless applications that require high throughput, extremely low latency, and deterministic data delivery. Its low-cost FPGA-based design also enables us to dramatically reduce cost and provide our customers with a higher-value product."

SCRAMNet GT is now available on the following platforms:

o PCI: GT200 PCI

o PMC: GT200 PCI

o VME: GT200 VME (6U)

Shared memory on each board starts at 128 Mbytes and supports data throughputs surpassing 200 Mbytes/sec.

SCRAMNet, which derives its name from Shared Common RAM Network, is CWCEC's multi-platform, replicated shared memory architecture. A SCRAMNet network uses a ring topology that supports up to 256 nodes. The resulting multi-processor system provides host-processor and operating system independence at each node. It's designed for use with real-time fiber optic communication systems and is optimized for the high-speed transfer of small data packets between computers when ultra-low latency data transmission is required. SCRAMNet GT's inherent low-latency performance supports the tight control-feedback loops required by many instrumentation and control systems for applications including man-in-the-loop simulators, range and telemetry systems, flight test, wind tunnels, engine test stands, robotics, particle accelerators, and industrial process control.

The use of shared memory has been adopted by industry in part due to its simple and intuitive user interface. After initialization, using function calls in the thin Application Program Interface (API), SCRAMNet GT appears to the user as standard memory. SCRAMNet GT supports Programmed Input/Output (PIO) for small data transfers. When larger data transfers are required the API provides function calls to support Direct Memory Access (DMA). Additional function calls are available to make the use of interrupts easy.