if (!isset($meta_desc)) { $meta_desc = "Leavitt Communications is a full-service international marketing communications and public relations agency established in 1991"; } ?>
News Release
More news releases by year:
2024
2023
2022
2021
2020
2019
2018
2017
2016
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
Return to: 2009 News Releases
TOKYO, Nov. 24, 2009 -- Imagination Technologies, a leading multimedia chip technologies company, announces the full roadmap for its Series2 generation of META processors, designed for the SoC-centric (System-on-Chip) age of silicon design.
The META family of 32-bit SoC processor IP cores is a unique range of embedded processors that combine both general purpose and DSP capabilities within a hardware multi-threaded execution infrastructure to deliver exceptionally high processor utilization and tolerance to SoC system latencies while also delivering new levels of real-time response that makes them ideal for SoC applications.
Indeed, the META processors are able to deliver both "hard real time" capabilities – the ability to respond to an external event in a single clock cycle including complete context switch – while also providing excellent Linux or Android applications processor platforms with fully coherent caches and MMUs. The META family stands out as a leading example of next generation SoC embedded processor architecture.
Furthermore, by the addition of features to facilitate highly efficient integration with multi-standard communications engines such as Imagination's ENSIGMA UCC programmable Wi-Fi and demodulation engines, META delivers a new generation of processors fully optimized for the internet and broadcast connected world.
SoCs being designed today place very different demands on embedded processors than those of a few years ago, as levels of systems integration continue to rise and applications make increasing use of high level APIs and operating systems. No longer is a single processor the highest priority user of system resources - it must increasingly share bus bandwidth, memory and other system resources with other sub-system processors for multimedia, communications and other specialized functions. This means the underlying architecture of embedded processors must be updated to reflect the fundamental changes in operating environment. It is with this evolving understanding of the practical realities of embedded processing in today's SoCs that Imagination has created the Series2 architecture for its META family of SoC processors.
Says Tony King-Smith, VP marketing, Imagination: "For too long chip designers have stuck to traditional concepts of CPU-based SoC design, which is increasingly not appropriate for the levels of system integration we're now seeing in SoCs. Embedded processors – whether running applications under Linux or highly optimized DSP algorithms - now need to be much more effective in ensuring every clock cycle counts to minimize power and maximize performance, Based on our years of experience of embedding META processors in many of our SoC IP cores and customer SoC designs, as well as our work with many of the world's most advanced SoC vendors integrating our high performance multimedia and communications engines with every major CPU architecture, we've built on the core strengths of our unique META processor architecture to meet the needs of SoCs far better than traditional CPUs for today's low power, high performance connected multimedia age."
There are now three META processor product lines:
META cores are capable of up to 1GHz operation on a 40G process, for a synthesized core. All the META cores are fully synthesizable using readily available standard libraries. Imagination is offering all members of the META Series2 family of SoC processors for licensing now.
META's hardware multi-threading delivers exceptional latency tolerance, and that translates in real terms to getting 2x or more usable processing power compared to conventional CPUs for similar silicon area, and often at lower clock speeds. The single cycle context switching combined with other advanced interrupt and low level hardware scheduling advances make META processors ideal for "hard real time" applications, where high data rates and demanding high speed event response mean single cycle context switch capability can make all the difference. This plus the ability to uniquely combine DSP and RISC-like thread configurations within a single unified engine allow designers to have a single embedded processor that delivers all the high level application as well as highly optimized DSP processing and real time performance demanded in many systems.
All META processors share common instruction sets, enabling easy migration from lower end cores to higher end as system designs progress from one generation to the next.,. This unique ability to use hardware multi-threading to create multiple virtual processors also offers SoC systems designers the ability to consolidate the various processing nodes used throughout the SoC under one consistent, silicon efficient processor architecture.
The META SoCket™ means that designers can simply switch one META core for another with little or no impact on the rest of the SoC design. Customers can start with a low or mid-range META, then follow a logical upgrade curve to higher levels of capability as experience with META and multi-threading grows and SoC architects see the benefits of utilizing META processors across multiple blocks in a SoC.
Many conventional processors are regularly stalled while waiting for memory requests to complete, typically causing as much as 50%, of the processing resource to be wasted. META's unique implementation of multi-threading enables it to change between hardware contexts each clock cycle, with up to four threads supported. This means that while a memory request for a Linux application on one thread is being resolved, an audio decoder can be progressed on another thread, data progressed though a communications protocol stack on another, and real-time hardware events serviced on another thread. Each thread sees the processor as its own, creating in effect up to four virtual processors utilizing a single data path. Caches can be configured to be coherent or as dedicated cache regions.
This unique approach to embedded processor design enables META to deliver up to twice the measurable throughput (e.g. Dhrystone) for the same clock speed and silicon area to comparable, state-of-the-art conventional processors. The scalability of the multithreaded META family delivers outstanding DMIPS for any given silicon area and clock-speed, typically ranging from 220 for META LTP to 1610 DMIPS for HTP (at 65nm). Compared to other processor solutions META can use a lower speed clock and less silicon area to deliver the same performance. Indeed, in production META can deliver a demonstrable 2.4 DMIPS per MHz. By utilizing Imagination's underlying architecture, which is built from the ground up to support hardware multi-threading, META processors are ideally suited to use in SoCs where high performance and sophisticated real time event handling are critical.
The META HTP family delivers the ultimate combination of powerful general purpose 32-bit processor with high performance DSP and low level control – all in a single, unified data path and cache. Capable of running many OS – and different OS on each thread if desired – the META HTP is the perfect solution for embedded SoC processor resources. Most configurations of HTP contain either 2 or 4 threads, with at least one configured for DSP and at least one thread used for full Linux. META HTP has optional double precision IEEE 754-2008 compliant floating point and 64-bit internal buses, loading a 64-bit word in a single cycle.
The META MTP family is designed to deliver the benefits of multi-threading and high performance DSP. Typically configured with two threads, META MTP processors have reduced ALU and cache resources to minimize die size, and are usually using an RTOS or native execution for applications. They are ideal for multimedia processing such as audio or communications. META MTP cores are used in several of Imagination's platform IP solutions and IP cores, where a mix of low level control and DSP or other mathematically intense processing tasks are required. Applications developed for a META MTP can usually be moved to larger META HTP processors without modification, offering an excellent upgrade path as applications mature and expand. META MTP has optional single precision floating point and 32-bit internal buses.
The META LTP is a single threaded 32-bit workhorse for deeply embedded control applications in SoCs. It is the ideal choice for low-level control and data flow processing, and is used extensively in Imagination's own IP cores.
Each thread in a META can be configured to be either to be 'GP (general purpose)' or 'DSP'. The GP threads utilize a highly optimized RISC-like instruction set, with smaller footprint instruction sets also available to minimize code size, and can also run DSP code. Each DSP-tailored thread also adds ALU resources and more registers to enable the processor to execute advanced DSP algorithms such as audio codecs, modems and more with maximum performance.
META processors run a wide range of operating systems, including full Linux (both single processor and SMP), and Imagination's own MeOS RTOS (Real Time Operating System). Indeed, META is the only processor architecture able to support multiple OS without the need for any virtualization layer – the underlying architecture inherently delivers this capability. META is highly suitable for porting other high level open source OS such as Symbian or Android. Threads running low level tasks can also be run natively without any OS for maximum efficiency. SMP Linux enables the benefits of multi-processing and hardware multi-threading to be brought together under a familiar Linux environment, enabling software engineers to take advantage of many of the benefits of hardware multi-threading without having to write any special code.
The META architecture throttles back aggressive code using AMA (Automatic MIPS Allocation) which also enables the system designer to adjust what percentage of total MIPS is made available to each thread. META can combine applications from legacy developments while minimizing the impact of 'selfish software' on other threads. META's super threading feature enables threads to run simultaneously provided they are not competing for the same resources, an approach with significant benefits over superscalar architectures. Super threading allows more work per clock cycle so the SoC can run at lower clock speeds while AMA handles system load-balancing to meet processing deadlines.
Imagination Technologies Group plc (LSE: IMG) – a global leader in multimedia and communication silicon technologies – creates and licenses market-leading processor cores for graphics, video, multi-threaded embedded processing/DSP and multi-standard communications applications. These silicon intellectual property (IP) solutions for systems-on-chip (SoC) are complemented by strong array of software tools and drivers as well as extensive developer and middleware ecosystems. Target markets include mobile phone, handheld multimedia, home consumer entertainment, mobile and low-power computing, and in-car electronics. Its licensees include many of the leading semiconductor and consumer electronics companies. Imagination has corporate headquarters in the United Kingdom, with sales and R&D offices worldwide. See: www.imgtec.com.
Return to: 2009 News Releases