Lexra Announces Breakthrough Network Processing Architecture NetVortex Scales from SOHO to OC192 Class Performance and Beyond for Next Generation Communication Infrastructure Development SAN JOSE, CA (June 14, 2000) - In communications switching systems OC48 data transfer rates represent a brick wall for traditional RISC processors and even some new so-called network processors. To break through this wall and provide processing power for OC192 and beyond, Lexra, a leading developer of processor cores for embedded applications announced today its NetVortex scalable network processing core architecture. It is the first processing core available for licensees to build solutions customized for applications ranging from cost sensitive Small Office Home Office (SOHO) Virtual Private Network (VPN) all the way to the backbone of tomorrow's OC192 data rate routers. NetVortex will be in production by Q4, 2000, and has been licensed to its first licensee.
"We are excited to bring out yet another highly-differentiated processor architecture optimized for this emerging market," stated Charlie Cheng, president and CEO of Lexra. "NetVortex will give Lexra's licensees the enabling technology, and a nine-month jump on its product development for a space where time-to-market is everything."
Network processors are the brains for communications, from the customer premise equipment all the way to the backbone of today's Internet. Traditionally, this function has been performed by custom developed ASICs, coupled with traditional RISC processor chips. However, the demand of communications switching systems for greater performance and increased functionality has heightened risk and stretched out the ASIC development cycle. At the same time, traditional RISC processors designed for yesterday's RISC workstations have failed to evolve to meet the rapidly changing needs of today's communications switching systems.
A new breed of specialized processor chips from Intel, IBM, and C-Port (now Motorola), are offering time-to-market advantages for off-the-shelf components. However, as components they lack the flexibility for system developers to add proprietary functionality, or achieve specific price/performance goals. The result is a loss of differentiation among system vendors. While many systems companies are incorporating these network processor chips in field trial systems, most believe that they will need to develop custom ASICs for production deployment.
NetVortex solves the performance/functionality vs. time-to-market dilemma for communications systems companies. First, its scalability allows designers to design in as much CPU performance as needed, and incorporate their own proprietary functionality into their ASIC designs.
Second, NetVortex speeds time-to-market because the designer can now start the project with much of the network processing architecture already completed. Like all Lexra products, the NetVortex is a microprocessor design database licensed to Lexra customers. As such, the licensee can incorporate the NetVortex into its own proprietary ASIC design. This approach enables the licensees to complete their ASIC design much faster, and still achieve unique technical differentiation.
The NetVortex Overview
NetVortex network processing architecture consists of two components: the LX8000 network engine CPU and the VortexBus. The LX8000 is responsible for performing the computation needed for network functions such as forwarding address calculation and repacketization. The VortexBus is responsible for moving packets from the network interface to NetVortex for computation, and then returning packets back to the network interface. In a target system design, up to sixteen (16) LX8000s, and up to four (4) VortexBuses can be used, thus delivering scalability for different applications.
At the heart of NetVortex is Lexra's LX8000 processor core. The LX8000 is highly optimized for network processing because it can handle up to eight different network data packets at any time. Packet transfer, as well as forwarding address calculation can take up many cycles, forcing traditional RISC CPUs to sit idle, waiting for the operations to complete. By handling up to eight packets simultaneously, the LX8000 is seldom, if ever idle. The result is a performance improvement of more than five-times the existing RISC technology, thus enabling a higher level of software functionality at lower cost and less power.
VortexBus is a unique design that differentiates NetVortex from traditional RISC processors. Traditional RISC chips incorporate system buses, optimized for Unix/Wintel desktop computing, but they are very inefficient for transferring network packets. As a result, the CPU often sits idle waiting for the packets to arrive. The VortexBus has a distinct protocol that is optimized for network traffic, making the entire bus dedicated to packet traffic. Furthermore, VortexBus delivers the packet traffic directly into the LX8000 memory without interruption or forcing the processor to wait during the transfer.
Lexra licenses NetVortex in two forms. The portable RTL version of NetVortex can be targeted to any silicon manufacturer. This version of LX8000 and VortexBus runs at 250 MHz worst case operating condition in a typical 0.15-micron silicon manufacturing process. The die area for each LX8000 is 2 mm2, and the power consumption is 100 mW.
The second version is a completed hard macro, optimized for a specific silicon manufacturing process, called SmoothCore. The SmoothCore NetVortex will run at 427 MHz in a typical 0.15-micron process. A single chip hardmacro implementation containing up to eight (8) LX8000 processors each with 16 Kbytes of instruction and 16 Kbytes of data memory can execute a large menu of OC192 services. This implementation provides 3600 MIPS of computing power and a VortexBus transfer rate of 119 Gbps. Furthermore, the solution can be contained in a mere 30 mm2.
Pricing and Availability
Customer shipments of the NetVortex will begin in 3Q 2000, with full production in 4Q 2000. The RTL version of the NetVortex is priced at $695,000 for one project. The SmoothCore version, Lexra's optimized hard macro of the NetVortex, is priced at $995,000 for one project.
Lexra, Inc. is a leading microprocessor developer specializing in RISC and DSP cores for the embedded market. In addition to competitive performance, small die size and low power consumption, Lexra's processor cores are also easy to use, easy to port and provide customers with cost effective solutions. Lexra is headquartered in San Jose, CA. Further information can be found at http://www.lexra.com.
Jonah McLeod Jr.
Director, Corporate Marketing
San Jose, CA
(408) 573-1890 x617
Senior Account Executive
Shelton Communications Group