Network / Packet Processing (NPP)
As network packet processing (NPP) applications demand greater and greater data bandwidths, especially with the advent of 100GigE, packet switching can become too complex for software alone to handle.
Applications must be able to dynamically manage packet processing and adjust to constantly changing security algorithms, and that’s where FPGAs are ideal. They provide a balance of flexibility while maintaining the performance necessary for these highly complex applications.
Why BittWare for Network Packet Processing:
- OpenCL support greatly simplifies FPGA development
- 100GigE, 40GigE, and 10GigE straight to the FPGA (no external PHY) for ultra low latency, saving hundreds of nanoseconds
- Complete server-based solutions available ready to use out of the box, integrated with multiple FPGAs, slot-cards, and software
- FPGA Development Kit with proven, hardware-specific IP
- FPGA IP for network or CPU acceleration from industry-leading partners
- Long standing FPGA platform vendor with a history of proven, successful, on-time supply
BittWare’s solutions provide high performance, high throughput and ultra-low latency FPGA platforms for financial and attached storage applications.
Network / Packet Processing… Built on BittWare.
CSPI/Myricom, an IT integration solutions and high-performance computer system provider, came to BittWare after determining that FPGAs were the logical solution for its next generation of network adapters. In a joint development effort, CSPI based its next-generation Myricom network adapter on BittWare’s Arria V GZ low profile PCIe board platform (A5PL), which provides two 40 GigE or eight 10 GigE network interfaces and optional DDR3 or QDRII+ SDRAM. The solution uses BittWare’s FPGA board platforms and value-add expertise, enabling Myricom to go to market quickly and efficiently with their proprietary Intellectual Property (IP). CSPI's family of network adapters integrates the company's application-targeted acceleration packages (DBL™ for financial trading and Sniffer10G™ for packet capture) and is optimized for highly accurate time stamping and low latency. Teaming with BittWare on the development of their Arria V based platform provided CSPI with a high quality foundation for their application-targeted IP.
NovaSparks, a leader in FPGA-based high performance and ultra-low latency market data solutions for the financial industry came to BittWare for an FPGA solution for their 3rd generation feed handler for financial applications. The feed handler for financial applications – NovaTick – is based on a BittWare Stratix V custom form factor board platform that features 8 QSFP+ and 8 SFP interfaces augmented by several QDRII+ SRAM chips and an optional DDR3 SDRAM module. The solution leverages BittWare’s FPGA board platforms and value-add expertise, enabling NovaSparks to go to market quickly and efficiently with their proprietary Intellectual Property (IP). Designed for time-sensitive trading strategies, NovaSparks' data feed handlers use the highest performance FPGA technology and board designs. Historically the company designed and developed these FPGA boards in-house. With its 3rd generation feed handler, NovaSparks chose to rely on BittWare to develop its new FPGA platform, allowing the company to deliver the best in class ultra-low latency market data solutions for the financial industry.
Network packet processing applications in the financial market must be able to process massive datasets within microseconds. FPGAs are ideal for these applications because they provide ultra-low-latency I/O, while also having much lower power requirements than GPUs or CPUs. Recognizing the value of FPGAs for providing high-performance access to the stock exchange, a well-known customer chose a BittWare FPGA board as part of their solution. The company went with BittWare’s S5PH-Q, which is a half-size PCIe card that provides low-latency serial I/O via front-panel QSFPs with direct access to a Stratx V FPGA.