Design Reuse

New Tensilica Processor Technology Optimized for Integration of Multiple Processors on SOCs


SAN JOSE, Calif.--(BUSINESS WIRE)--June 11, 2001-- Tensilica® Inc., the leading supplier of extensible, configurable processor semiconductor intellectual property (SIP), today unveiled its fourth-generation of the Xtensa(TM) processor technology. The Xtensa IV technology features advanced capabilities to facilitate the integration of multiple processors within System-On-Chip (SOC) implementations. In addition, the company has introduced a host of new configuration options and development tools that will allow system designers to further optimize the Xtensa architecture for specific applications.

``Many system designers acknowledge that the integration of multiple processors on SOCs is critical to next-generation embedded systems, but few processor vendors provide this capability beyond the simplest case: one CPU and one DSP,'' said Chris Rowen, president and CEO of Tensilica Inc. ``The Xtensa processor's flexibility has already enabled many Tensilica customers to design SOC devices with three or more Xtensa cores. The Xtensa IV core is even more multiprocessor-friendly, providing a development environment that enables designers to reduce time-to-market and optimize the performance of SOCs with multiple Xtensa processor cores.''

A New Definition of Multiprocessing

The traditional view of multiprocessing in general purpose computing systems has been a cluster of two, four, or eight or more identical processors coupled via traditional bus structures. Terms such as symmetric multiprocessing (SMP) or massively parallel computing have been used to describe these systems. By contrast, in consumer and communications SOCs, designers are deploying a combination of uniquely configured task engines -- many of which are tightly coupled to each other -- in dataflow or ``dataplane'' functions, often as an alternative to implementing fixed-function RTL logic.

``Tensilica's foray into CMP (chip-level multiprocessing) through its Xtensa IV architectural enhancements shows where SOC development is headed,'' said Steve Leibson, VP, Editorial Director, and Chief Analyst at MicroDesign Resources. ``The number of transistors available on commercially viable silicon is quickly reaching the point where it makes tremendous sense to place multiple processors on one die. A configurable core that can be optimized for a variety of tasks, like Xtensa IV, makes a tremendous amount of sense in these applications.''

Hughes Network Systems Chooses Xtensa for Multiprocessing SOC

Hughes Network Systems (HNS) recently chose the Xtensa processor for Spaceway(TM), the next generation satellite-based broadband network for consumers and enterprises. According to HNS, Spaceway's system architecture required a System on Chip containing multiple unique processors, each precisely tuned to an assigned function. ``After an extremely thorough evaluation of processors from several industry-leading companies, it was clear to us that Tensilica's Xtensa core provided the performance, time-to-market, and verification specifications that we required,'' said Dan Fraley, Hughes Networks Systems' senior vice president of engineering.

HNS will use numerous uniquely configured Xtensa processors in their SOCs for Spaceway. These will provide all of the processing requirements including high speed real-time data stream management plus running a leading RTOS on an Xtensa core configured as a high-end control processor.

Key Enhancements to the Xtensa Architecture

The Xtensa IV processor is the newest generation of the company's proven configurable and extensible microprocessor architecture. The Xtensa architecture provides a superior alternative to traditional rigid processors and simple configurable processor architectures that limit designers to the performance of either a ``one size fits all'' general-purpose MPU or a limited number of configurations. Tensilica's powerful, integrated hardware and software development environment offers thousands of configuration options and an unlimited range of customer-specific extensions which enable designers to carefully tune the processor for specific functionality. With an easy-to-use graphical interface, designers can take advantage of Tensilica's processor generator to create customized MPU solutions with specialized functions and instructions. Because these instructions are recognized as ``native'' by software, developers can simultaneously tune both application software and processor hardware to meet specific speed, power and feature goals.

The Xtensa IV core delivers all of the unique features of the Xtensa architecture and has been enhanced to also include the following:

System Simulation and Implementation

  • New System Modeling API: Tensilica has developed a new system modeling and simulation application programming interface (API) and function library. The API reduces overall design time by allowing rapid construction of C simulations of one or more Xtensa processors, on-chip memory, buses, bus bridges and other peripherals. As a result, designers can now explore topologies and load balancing for multiple processor systems and get detailed performance simulation with rapid iteration cycles.
  • C-Callable Instruction Set Simulator: Xtensa's Instruction Set Simulator has been enhanced to be C/C++-callable in multiprocessor system simulations. By integrating a configured instruction set simulator for each core produced by the Xtensa generator into an existing C simulation environment, Tensilica customers can rapidly create and iterate a C-level model of a multiple processor configuration running real compiled C-code.
  • Xtensa Local Memory Interface (XLMI): Xtensa IV offers a new high-speed local memory interface that can be used to optimize the performance of SOCs with multiple processors. The XLMI port can be used to tightly couple Xtensa processors used in dataflow instantiations, as well as integrate existing hardwired, high-performance complex state machine logic into the Xtensa processor's memory space. The XLMI interface can be configured up to a full 128-bit width, delivering up to 3.2GB/s (peak) low-latency memory bandwidth. The XLMI interface is supported by system modeling capabilities both in software, through Xtensa's Instruction Set Simulator, and in hardware, through bus functional models.
  • Robust Debugging Environment: Tensilica offers a robust, multiprocessing debugging environment that enables users to seamlessly transition from software simulation to actual hardware debugging using a consistent GNU-based software development tool chain. Xtensa IV adds selective hardware debug breakpoint capability for multiprocessing SOCs. Through a single daisy-chained, on-chip debug port, users can monitor and debug software using a single host with selective conditional breakpoints. On chip user-defined trace buffer functionality can also be added to the SOC using Xtensa's trace port for complete and comprehensive multiprocessing debugging solution.
Software Development Environment
  • New High Performance, Vectorizing C Compiler: Tensilica has expanded its C compiler offering to include the Xtensa C/C++ compiler. This new compiler generates optimized code to deliver a 15 percent performance improvement over the company's existing GNU C compiler solution. The new compiler is ANSI C++ compliant and is capable of automatic vectorization support for all configurations of Xtensa's Vectra(TM) DSP Engine. The company will continue to offer its existing GNU C Compiler as part of its standard licensing agreement.

  • Processor Configuration Options

    • Memory Management Unit: Xtensa IV offers a new memory management unit (MMU) option that supports separate data and instruction translation via look aside buffers. In addition, it supports a mixture of static and dynamic mapping for an optimal blend of functionality with a small footprint.
    • More Vectra DSP Engine configurations: Tensilica continues to enhance the flexibility of the Xtensa architecture through the addition of five new configuration options for the company's Vectra(TM) DSP engine. Vectra is a powerful DSP engine that has been optimized to handle high-performance DSP applications using fixed-point arithmetic. Vectra employs an efficient and easy-to-program vector architecture to deliver high throughput and low power dissipation for a variety of embedded SoC applications including communications, audio and imaging.
    • XT2000 Emulation kit: Tensilica now offers a second generation hardware emulation system with a larger capacity FPGA that enables designers to model more of the target system, including two or more Xtensa cores.
    Pricing and Availability

    The company's pricing structure is based on a licensing fee per design instance plus royalties based upon volume of processors manufactured. Licensing fees for a single processor configuration, including a complete, configured GNU-based software development toolchain, start at $350,000. The Xtensa C compiler, Xtensa Instruction set simulator, and Xtensa TIE Compiler are priced separately.

    Customers can begin taking advantage of Xtensa IV's new features in the third quarter. The standard license deliverables include source Verilog or VHDL RTL plus supporting EDA tool scripts, test suite, placement guidelines and the customized software tool chain.

    About Tensilica

    Tensilica was founded in July 1997 to address the fast-growing market for configurable processors and software development tools for high volume, embedded systems. Using the company's proprietary Xtensa Processor Generator, system-on-chip designers can develop a processor subsystem hardware design and a complete software development tool environment tailored to their specific requirements in hours.

    Tensilica's solutions -- now with over 30 licensees using the technology in more than 50 designs -- provide a proven, easy-to-use, methodology that enables designers to achieve optimum application performance in minimum design time. The Company has over 130 engineers engaged in research, development, and customer support from its offices in Santa Clara, California; Burlington, Massachusetts; Princeton, NJ; Houston, Texas; Oxford, U.K.; Stockholm, Sweden; Taipei, Taiwan, R.O.C.; and Yokohama, Japan.

    Tensilica is headquartered in Santa Clara, California (95054) at 3255-6 Scott Boulevard, and can be reached at 408/986-8000 or via www.tensilica.com on the World Wide Web.

    Editor's Notes:

    • ``Tensilica'' is a registered trademark and ``Xtensa'' and ``Vectra'' are trademarks belonging to Tensilica Inc. ``Spaceway is a registered trademark of Hughes Electronics Corporation. All other registered trademarks or trademarks are property of their respective owners.
    • Tensilica's announced licensees are, in alphabetical order, Berkeley Wireless Research Center, Broadcom, Cisco Systems, Conexant Systems, Fujitsu Limited, Hughes Network Systems, Marvell/GalileoTechnology, Mindspeed Technologies, National Semiconductor, NEC Networks, NEC Solutions, NTT, ONEX Communications, Osaka and Kyoto Universities, TranSwitch Corporation and ZiLOG.
    • Visit Tensilica at DAC 2001, June 18 - 21, Las Vegas Convention Center, booth number 3739.

    Contact:
         Tensilica, Inc.
         Stephen Roddy, 408/566-1748 (Reader)
         roddy@tensilica.com
         Kim Alfaro, 408/327-7343 (Editorial)
         kim@tensilica.com
              or
         Tanis Communications
         Joany Draeger, 650/365-3395 (Editorial)
         joany@taniscomm.com

     

    Previous Page