Achronix introduces FPGA memory bandwidth, PCIe acceleration board

Achronix Semiconductor Corporation has launched a new PCIe form factor Accelerator-6D accelerator board, said to be the industry’s highest single-FPGA memory bandwidth, PCIe add-in card for high-speed data centre acceleration applications. The board integrates a Speedster™22i HD1000 FPGA with 700,000 look-up tables that connects to six independent memory controllers allowing for up to 192 GB of memory and 690 Gbps of total memory bandwidth.

  • Wednesday, 22nd June 2016 Posted 8 years ago in by Phil Alsop
Each DRAM controller on the HD1000 runs at 1,600 MT/s and connects to two SODIMMs allowing for single, dual- and quad-rank SODIMM and SORDIMM DDR3 operation. The board also has four QSFP+ connectors for 10G/40G Ethernet connectivity and supports PCIe Gen3 ?8 operation. The controllers for the DDR3, Ethernet and PCIe interfaces are implemented as embedded hard blocks inside the HD1000 FPGA, which eliminates the requirement to use valuable programmable resources inside the FPGA to implement these functions. In addition, the embedded hard controllers guarantee timing on these complex high-performance interfaces, enabling designers to focus their valuable time on developing data centre acceleration applications.

The Accelerator-6D board is the only FPGA-based PCIe board that has 6 independent DRAM memory ports connected to a single FPGA. Each independent port can be configured with up to 32 GB of DDR3 memory. The memory ports combined with the 4 QSFP+ modules that support 4x 40G Ethernet is the perfect platform for data centre architects to develop intelligent NIC cards for NFV network acceleration or network security.

“The Accelerator-6D board offers a unique value proposition to the data centre, high-performance compute market segment that uses PCIe add-in cards,” said Steve Mensor, Vice President of Marketing, Achronix Semiconductor. “The Accelerator-6D board offers the highest memory bandwidth for an FPGA-based PCIe form factor board, and memory bandwidth is typically the bottleneck of high performance computer systems.”