Thursday, 13 January 2011

Low profile PCI Express platform delivers deterministic, low latency, high speed memory sharing

GE Intelligent Platforms Announces Reflective Memory Node Card For Demanding Simulation, Process Control And Data Acquisition Applications

PCIE-5565PIORC low
profile PCI Express reflective
memory node card
GE Intelligent Platforms has announced the PCIE-5565PIORC low profile PCI Express reflective memory node card, further extending the company’s RFM 5565 family of reflective memory solutions and taking advantage of GE’s extensive experience in the technology. Designed for real-time applications such as simulation and training, industrial process control and data streaming and acquisition, the low profile PCIE-5565PIORC will also serve a broad range of commercial, telecommunications and military/aerospace applications that require redundancy for high availability. Such applications are characterized by the need to move and share data at fast, deterministic rates and, in many cases, the only solution in such environments is reflective memory.

A reflective memory network is a special type of shared memory system which enables multiple, separate computers to share a common set of data. Reflective memory autonomously replicates the contents of one processor's memory to the memory nodes of all other network members.

“In environments where determinism, simplicity of implementation, zero software overhead and operating system independence are significant factors, reflective memory provides a unique solution,” said Wayne McGee, Leader, Commercial Product Management at GE Intelligent Platforms. “While other technologies – such as Gigabit Ethernet, Infiniband and Fibre Channel – were predicted to displace it, that hasn’t happened, as they are not designed to deliver the sheer performance required. Reflective memory, on the other hand, is designed specifically to enable the sharing of data among processors in real time by mission-critical applications.”

The PCI-5565PIORC is characterized by its low profile implementation which is a vital requirement for many customer implementations. The board features 2.12 Gbaud serial connection speed, improved Programmed I/O (PIO) read performance, field upgradeable firmware and RoHS compliance, and the maximum PCI burst rate has been doubled to 512MBytes/second: larger memory (256MBytes) is also supported. GE’s reflective memory networks can be scaled to 256 nodes and operate at speeds up to 174 Mbytes/second. Distance between nodes can be up to 10 kilometres.

Reflective memory can offer an ideal solution in many demanding applications because of its low latency and determinism. Using reflective memory, the typical node-to-node propagation time is on the order of 1 microsecond. Thus, in a system of 30 nodes, it would take only 30 microseconds to propagate through the network to all nodes. Achieving the same latency using other network technologies, such as Ethernet, is extremely difficult – if not impossible - even using datagram broadcast because of IP protocol overheads, addressing, and memory write times.

No comments:

Post a Comment