CLOUD
MIPS Tech Targets Multi-CPU SOC Designs With New 32-Bit Processor Core
- Written by: Writer
- Category: CLOUD
SAN JOSE, CA -- MIPS Technologies, Inc. (Nasdaq: MIPS, MIPSB), a leading provider of industry-standard processor architectures and cores for digital consumer and networking applications, today announced a new 32-bit synthesizable core designed to optimize SOC designs implemented with multiple CPU cores. The emerging trend in multi-CPU SOCs addresses the rapidly increasing bandwidth requirements in next-generation broadband and networking devices. The new MIPS32(TM) M4K(TM) core gives designers higher performance and greater flexibility to achieve higher system throughput while controlling silicon cost. The flexibility and re-programmability enables upgrades in software as protocol specifications or market requirements evolve. Applications for the M4K core include data plane processing, as well as deeply embedded control processors, networked storage, residential gateways, set-top boxes and smart mobile devices. "Multi-CPU SOCs are required to meet the high-bandwidth demands of advanced networking equipment," said Linley Gwennap, principal analyst of The Linley Group. "Modern IC process technology easily supports many CPUs on a single chip; the problem lies in connecting and debugging such complex designs. The M4K core provides an efficient and effective solution to this problem while maintaining compatibility with the industry-standard MIPS instruction set and tool chain." The M4K core features a typical clock speed over 300 MHz, yet minimum power consumption is only 0.10 mW/MHz, and core size is as small as 0.3 mm2 in 0.13-micron processes. It features code compression to reduce memory size, and it is the first core to utilize the MIPS32 architecture enhancements announced last October. These include bit field instructions for easier handling of packet information, support for vectored interrupts to decrease interrupt latency, and multiple register sets for faster context switching. Multi-CPU designs, in particular, benefit from the core's high-speed cacheless SRAM interface, user-defined instruction-set extensions to create highly differentiated features and optimize performance, and support for easy multi-CPU simulation and debug. For networking applications, the M4K core is code-compatible with MIPS-based 64-bit processors in the control plane, which gives networking system engineers more flexibility to allocate functions performed by the data plane and control plane processors in order to boost processing efficiency. Features of the synthesizable M4K core include: -- Typical performance: 300 MHz, 405 DMIPS (0.13-micron generic process) -- Optimized cacheless SRAM memory interface that enables deterministic performance and reduces die size -- 5-stage pipeline that allows most instructions to execute in one cycle -- Power consumption as low as 0.10 mW/MHz -- Core size as small as 0.3 mm2 -- MIPS16e(TM)code compression that reduces memory requirements by as much as 40 percent -- Packet manipulating bit instructions for packet header and deep-packet examinations and editing -- Vectored interrupts that reduce latency -- 1, 2 or 4 general-purpose register sets for fast context switching -- Fast multiply/divide unit -- Enhanced JTAG (EJTAG) with PC and data trace support for easy multi-CPU debugging -- Seamless, upward compatibility with MIPS64(TM)-based cores "The M4K core is the first core based on an industry-standard architecture to have user-defined instruction-set extensions. It is also the first to include the enhanced MIPS32 architecture, which offers faster and more flexible packet processing and low-cost interrupt handling," said Kevin Meyer, vice president of marketing at MIPS Technologies. "The M4K core also comes with a robust multi-CPU design environment that enables SOC designers to optimize their designs and quickly bring them to market."