BDTI’s DSP Insider Archives
|
||
BDTI’s DSP Insider Archives |
|
|
| HOME << |
|
|
|
||
This month:
Intel Doubles PXA2xx’s Signal Processing SpeedToday Intel announced the PXA270, PXA271, PXA272, and PXA273—the latest members of its PXA2xx family of application processors. These new PXA27x family members are the first processors to include the “Wireless MMX” extensions to Intel’s ARM-compatible XScale architecture. The Wireless MMX extensions enable the PXA27x to perform 64-bit wide single-instruction multiple-data (SIMD) operations. For example, the PXA27x can perform four 16-bit multiply-accumulate operations per cycle. In comparison, older PXA2xx family members can complete a maximum of two 16-bit multiply-accumulate operations per cycle. (For details of the Wireless MMX extensions, see the October 2002 DSP Insider.) The PXA27x also operates at a higher clock rate than its predecessors. While older PXA2xx family members operated at a maximum clock rate of 400 MHz, the PXA27x operates at up to 624 MHz. The combination of higher parallelism and higher clock rate makes the PXA27x significantly faster than its predecessors: based on an analysis of BDTImark2000™ scores, the 624 MHz PXA270 is over two times faster than the 400 MHz PXA255 on signal-processing tasks. (See http://www.BDTI.com/bdtimark/BDTImark2000.htm for these and other benchmark scores.) The PXA27x also offers expanded memory integration. First, the PXA27x adds 256 Kbytes of level-two SRAM to the 66 Kbytes of cache memory found on earlier PXA2xx family members. The PXA27x also expands the amount of “stacked” memory. Earlier PXA2xx family members contained up to 32 Mbytes of flash memory stacked with the processor in a multi-chip package. In comparison, the PXA271 contains both 32 Mbytes of stacked flash and 32 Mbytes of stacked SDRAM, and the PXA272 contains 64 Mbytes of stacked flash. The PXA27x also provides a richer set of on-chip peripheral interfaces than its predecessors. For example, the PXA27x adds a camera interface that can connect to a variety of CMOS and CCD cameras.
According to Intel, the PXA270 will operate at up to 624 MHz, while
other PXA27x family members will operate at up to 520 MHz. The 312
MHz PXA270 costs $32.00 in 10,000-unit quantities. As of this
writing, pricing for the other family members was not available. All
versions of the PXA27x are available in sample quantities now.
According to Intel, the PXA27x will enter full production this
quarter.
Embedded C Adopted—and ExplainedLast month the ISO officially adopted Embedded C, an extension to the C programming language that will ease signal-processing software development. The ISO will publish the Embedded C specification as a “technical report.” An ISO technical report is similar to a standard, but carries somewhat less authority. Two features of the Embedded C specification are particularly relevant for signal-processing applications. First, the specification supports fixed-point data types, which are often used by signal-processing software. Second, the specification supports memory space qualifiers, which allow programmers to make effective use of the multiple memory spaces found on most DSP processors. (See the September 2003 DSP Insider for details on these features.) Numerous companies have offered proprietary variants of C with these and related features for many years. However, these proprietary solutions take varying—and incompatible—approaches to supporting signal-processing applications. In contrast, Embedded C has the potential to enable code portability across processors and to promote a consistent programming style for signal-processing applications. One company offering a proprietary solution, ACE Associated Compiler Experts bv, initiated the ISO Embedded C effort in 1998 when it proposed that its DSP-C language be adopted as an extension to the C programming language. ACE recently announced that the ISO adopted DSP-C as part of the Embedded C specification. While Embedded C does include features found in DSP-C, the ISO specification differs from ACE’s original proposal in important ways. In addition, Embedded C includes several sets of features that were not in the original DSP-C proposal. These changes and additions to ACE’s original proposal reflect the contributions of numerous individuals and companies. The Embedded C specification has been adopted at a time when signal-processing software development has largely migrated to C. It will be interesting to see whether compiler developers will fully comply with the Embedded C specification, or if incompatible variations called “Embedded C” will proliferate.
The Embedded C technical report is available online at
http://std.dkuug.dk/JTC1/SC22/WG14/.
BDTI Case Study
This month: Measuring Multimedia PerformanceAt first glance, measuring processor performance on multimedia applications can seem straightforward. Many multimedia applications are based on published standards and widely available software. For example, MPEG-4 video decompression software is available for most popular processors. Because such software is often readily available, measuring multimedia performance may seem to be a simple matter of checking the processor vendor’s published performance data for the relevant software modules. In reality, reliable measures of processors’ multimedia performance are hard to obtain. A number of factors confound attempts to obtain meaningful, apples-to-apples performance comparisons. One of these is the data-dependent processing loads posed by many multimedia tasks. For example, the processing power required for MPEG-4 video decoding can vary by as much as a factor of two depending on the content of the video. Further complicating matters, vendors may omit key application components from their performance quotes. For example, MPEG-4 video decoding is nearly always followed by color space conversion, but vendors rarely include color space conversion in their performance figures. This is a crucial omission, because color space conversion is computationally intensive. On an ARM9E running a typical MPEG-4 decoder at 176x144 pixels and 15 frames per second, for example, color space conversion requires roughly the same computation power as MPEG-4 decoding. Navigating these complexities requires in-depth knowledge of multimedia applications, processors, and software. Thanks to its years of experience in each of these areas, BTDI has the expertise needed to obtain meaningful, comparable measures of processors’ multimedia performance. In one recent project, BDTI compared processors for a variety of mobile multimedia applications. For each application, BDTI analyzed the performance requirements of the resource-intensive tasks such as video compression. BDTI used these requirements to evaluate the processors’ multimedia processing capabilities, including the capabilities of their hard-wired multimedia accelerators.
BDTI will make its multimedia performance evaluation expertise widely
available when it releases its new multimedia benchmark, currently
under development. To learn more about this forthcoming benchmark and
BDTI’s other multimedia evaluation capabilities, contact Jeremy
Giddings at giddings@BDTI.com.
Impulse Response, by Jeff Bier
Multiprocessor Migraines?Once upon a time, most signal processing applications were powered by single-processor chips. Today, though, there is an increasing trend towards using complex, heterogeneous multiprocessor chips. One such chip is Texas Instruments’ recently announced OMAP 2, which contains a microprocessor core, a DSP core, and multiple application-specific coprocessors. The primary motivation for this shift is not processing speed; the fastest uniprocessor DSPs and general-purpose microprocessors are fast enough for many performance-hungry signal-processing applications. Instead, multi-processor chips are gaining popularity because these chips deliver not only strong computational speed but also prices suitable for high-volume applications and (at least in some cases) energy efficiency suitable for battery-powered systems. The obvious down-side of heterogeneous multi-processor chips is that they are more complex. Instead of a single processor architecture and a single tool chain, users have to contend with learning multiple architectures and tool chains. Many chip vendors attempt to ease this process by providing software and middleware so that the user doesn’t have to develop all of the code from scratch. Nevertheless, users will likely need to develop, optimize, and debug some software for each processor on the chip. Perhaps even more daunting is the task of partitioning an application across different processors. Unlike “channelized” applications, which often use multiple instances of the same processor to execute the same software in parallel, applications that use heterogeneous multi-core chips run different portions of the application on different processors. So which portion goes on which processor? Ideally, you’d like to run each block of code on the processor that’s most efficient for that block. But it may not be obvious which processor is most efficient until you’ve actually implemented and optimized each block on each processor—which is not a viable approach. In addition, you’ll have to consider the overhead associated with inter-processor communication—that may mean that certain blocks are best mapped to a less-efficient processor to avoid excessive shuttling of data.
When heterogeneous multi-processor chips provide significantly better
performance, energy efficiency, or cost than single-processor
alternatives, system developers will put up with some added
complexity. But it is up to the chip vendors to provide the needed
infrastructure to help make that complexity manageable.
BDTI Goes Inside the new Intel PXA27xThe Intel PXA27x—announced today—is a 32-bit fixed-point embedded processor family that targets PDAs and mobile wireless products—most notably, smart phones. The PXA27x is the next generation of Intel Personal Internet Client Architecture (PCA) application processors. Like all BDTI Inside reports, Inside the Intel PXA27x delivers results from benchmarking with the BDTI Benchmarks™ as well as comparisons of the PXA27x to competitor architectures. Inside the Intel PXA27x will help you understand how the PXA27x performs on key signal processing functions and will provide insights into this new architecture.
For more information or to order the report, go to
http://www.BDTI.com/products/reports_pxa27x.html.
BDTI’s Buyer’s Guide for 2004 Provides Analysis and InsightBuyer’s Guide for 2004 includes new benchmark results for ADI’s TigerSHARC and Blackfin processors and updated benchmarks for ADI’s SHARC, Motorola’s MSC8101 (StarCore), and TI’s ’C55x and ’C64x. The report provides updated analysis of products in each processor family, including speeds, prices, power consumption, and peripherals. Other new features to this year’s edition include the introduction of the BDTImemMark2000™ , a single-number metric showing overall memory efficiency, innovative “radar” charts that summarize processor performance, a new layout, and new formatting. The 2004 Buyer’s Guide is 584 pages in 8.5 x 11 inch format, spiral bound for easy desk use. The first copy is $2,695, including shipping via FedEx to North American addresses (for international shipping, add $75). Additional copies are discounted substantially.
For previews of the report and more information, go to
http://www.BDTI.com/bg04.
Join BDTI at Embedded Processor Forum 2004In-Stat/MDR’s annual Embedded Processor Forum brings together key players from the world of embedded processing for four days of seminars and conferences. Conference sessions are highlighted by groundbreaking new product announcements from established leaders and up-and-coming innovators in processor technology. This year, BDTI applies its expertise and insight to create a conference session focused on one of the hottest areas for processor technology—video. Join Jeff Bier, BDTI’s general manager, on Wednesday, May 19 for “Processors for Video Applications.” This session will include announcements and detailed presentations from MobilEye, Ultra Data Corp., Cradle Technology, and Texas Instruments.
For links to more information about Embedded Processor Forum, visit
http://www.BDTI.com/bdti_whatsnew.html#epf.
Call for Papers: GSPx 2004GTC is seeking proposals for papers to be delivered at the Global Signal Processing Expo and Conference 2004 to be held September 27-30, 2004 at the Santa Clara Convention Center. The deadline for submitting abstracts is April 30, 2004.
The conference will bring together leading developers and design
engineers presenting over 700 peer-reviewed papers. In addition, over
150 of the industry’s leading embedded signal processing companies
will be featured at the trade show. More information, including the
GSPx call for papers, can be found on the event website at
http://www.gspx.com.
BDTI Updates Benchmark ScoresBDTI has released new BDTImark2000™ scores for the Intel PXA27x and the Analog Devices ADSP-TS201 “TigerSHARC” and updated the score for the Texas Instruments TMS320C67x. The new ADSP-TS201 score measures the processor’s performance on 32-bit floating-point arithmetic. This score complements the previously released score that measures the ADSP-TS201’s performance on 16-bit fixed-point arithmetic. For these and other scores, go to http://www.BDTI.com/bdtimark/BDTImark2000.htm.
The BDTImark2000™ is a summary measure of digital signal processing
speed distilled from a suite of DSP benchmarks developed and
independently verified by BDTI.
About BDTIBDTI is an independent source for digital signal processing technology analysis and optimized software development services. From rigorous technical analyses of processors for DSP, such as the Inside series of processor analyses, to highly regarded technology seminars, BDTI is the trusted independent source for reliable information on digital signal processing technology. As a software developer, BDTI is known for highly optimized implementations of signal processing algorithms and applications and for solutions to complex problems of integration, code size, and performance.
For more information, visit our Web site at http://www.BDTI.com.
As previously announced, a new newsletter, Inside DSP, published jointly by CMP Media’s EE Times and BDTI, will soon take the place of the DSP Insider. Both newsletters are free. If this newsletter was forwarded to you and you would like to receive the new Inside DSP newsletter regularly, register at http://www.BDTI.com/dspinsider.htm. If you do not wish to receive the new BDTI-CMP Inside DSP newsletter, send an email message to dspinsider@BDTI.com with the words “Remove me” in the subject line.
Previous issues of BDTI’s DSP Insider are archived and will continue
to be available at http://www.BDTI.com.
BDTI’s DSP Insider © 2004 Berkeley Design Technology, Inc. |