Embedded Goes ParallelBe the first to comment!
Embedded designers have traditionally made scant use of “extra” cores of Atom and other chips. The exception: offloading easily parallelized, compute-intensive tasks, such as background data compression of large files before storage.
But strong consumer demand for smarter devices and wider proliferation of parallel processing techniques is changing all that.
Embedded designers are increasingly making use of sophisticated parallel programming methods and tools previously reserved for high-performance computing (HPC).
Software toolkits for embedded and mobile systems including Intel’s System Studio can take advantage of multicore processors at all levels of system design, from the on-chip registers to the operating system to the applications and services. Source: Intel
“Even on the smaller embedded processors today, multi-cores are becoming common,” noted Herb Hinstorff, Intel’s director of marketing for developer products at the recent Global Press eSummit 2013 (April 15-17, Santa Cruz, Calif.)
As a result, he said, programmers need the same parallel compilers, debuggers, libraries, performance analyzers and thread checkers traditionally used for HPC. “In this new ‘Internet of Things’ age, even simple mobile devices today are getting more intelligent, driving the need for the security, connectivity and other capabilities that can make good use of parallel processing,” Hinstroff said.
Intelligent embedded systems in consumer, medical, automotive, retail and other diverse areas are being linked together wirelessly by cloud resources. Source: Intel
To meet those needs, Intel’s developer efforts — spearheaded by more than 14,000 software group employees –have started integrating many of the same parallel processing capabilities in Intel Parallel Studio and Intel Cluster Studio into its embedded development suite called Intel System Studio. To date the company claims more than 12,000 embedded design wins.
“What we have done is taken the parts of Parallel Studio and Cluster Studio relevant to embedded systems and built them into System Studio, “ Hinstroff explained, “integrated with all the other tools that embedded designers need, such as longer support lifetime, dedicated channels for embedded, system trace, JTAG [Joint Test Action Group] and other capabilities embedded designers need to get their products to market quickly.”
Intel’s System Studio for embedded devices, from smartphones and tablets to the wider range of Internet of Things, have had the appropriate parallel-processing capabilities of Parallel- and Cluster-Studio built-in. Source: Intel
Intel System Studio includes parallel compilers, debuggers, libraries plus special tools to analyze an entire embedded system-on-chip (SoC). This aids performance, power consumption and reliability important for mobile consumer devices.
For instance, Intel System Studio debuggers not only identify memory and threading errors, but provide a system-wide view of every register on an SoC. Doing so lets designers identify, for example, synchronization problems that could cause an SoC to fail in the field if not detected during development.
Intel System Studio compilers take advantage of multicores to boost performance. Likewise, its libraries are designed to scale well from single- to multi-thread solutions. Intel System Studio supports all conventional operating systems such as Linux and Windows, along with real-time-operating systems (RTOS) popular for the smallest embedded systems, including Wind River‘s (an Intel subsidiary).
April 22, 2013 R. Colin Johnson, Slashdot Contributing Editor