VMIACC-5595 Reflective Memory HUB

tjthkj

VMIACC-5595 Reflective Memory HUB [Copy link]

Reflective memory price, PCI5565, PCI-5565, PMC5565, VMIC5565, Reflective memory network, Reflective memory card, GE Reflective memory

Reflective memory network can greatly improve the communication speed between cluster systems, so it is particularly suitable for replacing applications where traditional LANs, bus repeaters and DMA cannot meet the needs. The 5565 series reflective memory card newly launched by GE FUNAC has a transmission speed of 174M bytes/second. The use of optical fiber can not only connect more nodes (up to 256 nodes), but also has high impact and anti-interference performance. Reflective memory card is a completely physical communication and does not require any software protocol overhead, which is unmatched by other LAN transmissions. Test results show that there is less than 400 nanoseconds of delay from the time data is written to RAM to the time it is transmitted to the reflective memory card of another node.

The reflective memory works as a dual-port memory, and the local host writes to its reflective memory address space, which is a port of the local memory. The RFM board automatically transmits this new data from its other port, which is an optical fiber connected to the network of the ring architecture, working at a baud rate of 2.1G. The next RFM board in the network receives this new data, and its local memory will be updated within 400ns. The reflective memory network can greatly save the cost of software development because there is no need to write and test any application code, nor to write corresponding documents, and there is no need to maintain assembly information and decoding information, as well as parse data from input information. Software costs are the most expensive independent expenses that a company can control. RFM products do not require software and learning curves, which will greatly save product time to market.

Reflective Memory Networks offer many features beyond standard networks. Features such as dual-ended RFM, high-speed data transfer, and software transparency make RFM product networks easy to use and a powerful solution for multi-computer connectivity. Compared to traditional connectivity methods that require additional software development time, testing, maintenance, documentation, and additional CPU requirements, RFM product networks offer a high-performance, cost-effective option.

The interface board uses FPGA (field programmable gate array, users can define the functions of FPGA chips by writing hardware description language programs, which can realize various complex combinations and sequential logic circuit functions, as well as various CPU, embedded processor and DSP functions, with high speed and full parallelism) chip as the control core. FPGA controls PCI9054 chip to exchange data with the computer PCI interface. The serial optical signal on the optical fiber network is converted into a serial electrical signal through the optical transceiver module. The serial electrical signal converts the 2.125GHz serial data into 16-bit 106MHz parallel data through the serial/deserializer (TLK2501). FPGA receives parallel data from the serial/deserializer TLK2501, analyzes the data, detects the data frame, and forwards the data frame to be forwarded to the serial/deserializer (TLK2501) in time for parallel-to-serial conversion according to the data frame information. At the same time, FPGA writes the received data into SDRAM and reads the data from SDRAM according to computer instructions.

Considering the transmission bandwidth, anti-interference, equipment layout and other aspects, using the fiber optic reflective memory interface board for data transmission between device interfaces is a good transmission solution. This can make the equipment layout in the test process more flexible, adapt to the evolving computer data transmission needs, and meet its reliability, real-time, deterministic and high-speed data transmission requirements. The fiber optic reflective memory interface board forms a communication system through optical cable connection, with low latency, large data transmission capacity and convenient wiring. The fiber optic reflective memory interface board adopts a distributed shared memory strategy. Data transmission on the network does not occupy local computer CPU resources, and does not use multi-level complex data transmission protocols, thereby achieving the purpose of improving data transmission real-time and reliability.