BDTI’s DSP Insider Archives
|
||
BDTI’s DSP Insider Archives |
|
|
| HOME << |
|
|
|
||
This month:
Analog Devices DRAMatically Improves TigerSHARCAnalog Devices announced three new TigerSHARC family members, the TS201, TS202, and TS203, at last month’s Embedded Processor Forum. These new parts represent a major upgrade to the TigerSHARC family in terms of speed, memory capacity, and cost effectiveness. The new TS20x family members operate at up to 600 MHz—twice the clock rate of the older TS101. This doubling in clock speed closes the performance gap between TigerSHARC and one of its key competitors, Texas Instruments’ ’C64xx. BDTI is currently evaluating the TS20x using its BDTI Benchmarks™. Preliminary results suggest that in terms of overall 16-bit signal-processing speed, a 600 MHz TS20x is about as fast as a 720 MHz ’C64xx. If the TS20x is playing catch-up in the speed contest, it has taken a clear lead in terms of memory capacity. The TS20x includes up to 3 Mbytes of on-chip memory. In comparison, both TI’s ’C64xx and Intrinsity’s FastMATH processors offer just over 1 Mbyte of on-chip memory. Competing PC CPUs lag even farther behind: both Motorola’s MPC74xx and Intel’s P4 top out at roughly half a Mbyte of on-chip memory. The TS20x is also far more cost-effective than the TS101. For example, a 300 MHz TS101 with 768 Kbytes of on-chip memory costs $199. In contrast, a 500 MHz TS202 with 1.5 Mbytes of on-chip memory costs $149. Thanks to its lower prices and larger memories, TigerSHARC is now more competitive with other DSPs. For example, it is roughly aligned with the ’C64xx in terms of speed per dollar and memory per dollar. ADI says it was able to lower prices even as it increased memory capacity by switching on-chip memory from SRAM to embedded DRAM (eDRAM). According to ADI, a TS20x with 3 Mbytes of eDRAM occupies about the same die area as a TS101 with 768 Kbytes of SRAM. ADI says the switch to eDRAM has other benefits as well, such as reduced power consumption. Although the TS20x is not the first embedded processor to use eDRAM—see the November 2002 edition of the DSP Insider for other examples—it is the first mainstream DSP to use this memory technology.
The TS201, TS202, and TS203 are sampling now at 500 MHz. The TS201 is
also sampling at 600 MHz. Volume production of the TS20x is expected
to begin in the second quarter of 2004. Pricing for the TS20x ranges
from $35 for a 500 MHz TS203 with 512 Kbytes of eDRAM to $299 for a
600 MHz TS201 with 3 Mbytes of eDRAM. (All prices quoted in this
article are for 10,000-unit orders.)
NeoMagic Unveils Unusual Application ProcessorA central theme of last month’s Embedded Processor Forum was the challenge of combining high speed with low power consumption. NeoMagic presented an unusual approach to this problem in its “MiMagic 6” application processor. (For more information on application processors and an overview of the competitive field, see the March 2003 edition of the DSP Insider.) Most application processors contain DSP coprocessors and/or hard-wired accelerators for video and image processing. Texas Instrument’s OMAP161x family, which contains a DSP coprocessor and video accelerators, is a prime example of this approach. In contrast, MiMagic 6 uses a programmable, highly parallel “Associative Processing Array” (APA) for compute-intensive media processing. The APA is a 512-row by 160-column array of 1-bit elements. Each element functions both as a processing element and as a 1-bit memory element. Similarly, the entire array can be thought of both as processing engine and as a 10 Kbyte cache. The APA performs operations in a column-wise fashion, with each operation affecting all 512 rows in parallel. Native APA operations are limited to compare, write, and move. More complicated operations can be constructed from sequences of these basic operations. For example, the APA can perform 16-bit addition using a 48-cycle sequence of operations. Because the APA operates on all 512 rows at once, this 48-cycle sequence translates into a throughput of about ten 16-bit additions per cycle. Consequently, a 100 MHz APA can perform about 2.5 times more 16-bit additions per second than the 204 MHz DSP core in the OMAP161x. The MiMagic 6 has the potential to achieve good energy efficieny. As illustrated in the preceding example, the MiMagic 6 may be able to save energy by operating at lower clock speeds than competitors like the OMAP161x. In addition, the APA’s tight integration of storage and compute resources may enable it to expend far less energy on memory accesses than competing architectures. However, the APA will likely be more difficult to program than competing DSP architectures. Fortunately, NeoMagic plans to provide APA software implementations of key functions including MPEG-4, speech compression, and 3D graphics primitives. NeoMagic also plans to provide libraries of basic word-level operations like add and multiply. MiMagic 6 might have performance advantages over its competitors, but its success will likely depend as much on the quality of these libraries as on any hardware superiority.
The MiMagic 6 is expected to begin sampling in the third quarter of
2003. It is priced at $18 in 10,000-unit quantities.
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 easily vary by 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 time 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 later this
year. 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
The Consumer Audio RenaissanceConsumer audio products have always held a special fascination for me. Indeed, it was my childhood interest in audio gear that largely motivated my pursuit of an engineering career. But during most of my career, mainstream consumer audio technology has been pretty staid territory, with little in the way of exciting technology and few compelling new products. More recently, though, I’ve been delighted to see what I believe is the beginning of a revolution in consumer audio equipment. The emergence of low-cost chips with heavy-duty digital signal processing capabilities is making it possible to bring new levels of quality, convenience, and functionality to consumer audio products. Some of the most interesting work in consumer audio is focused on sound reproduction: specifically, using sophisticated signal processing to overcome limitations in listening spaces and loudspeakers. At the recent Audio Engineering Society (AES) conference in Denmark, several companies showed off exciting innovations in this area. Bang and Olufsen, for example, demonstrated high-end home audio loudspeakers that compensate for imperfections in a room’s acoustics. When these speakers are first installed in a room, they use sensors to gather information about the room’s acoustical properties. Processors in each speaker then run filter-design software to create filters that compensate for defects in the room’s response. The resulting filters process audio signals before they are sent to the speakers’ built-in amplifiers. In this way, the speakers eliminate certain unpleasant acoustical effects, such as the “boominess” that can occur at low frequencies when a speaker is placed in the corner of a room. Today this capability is available only in B&O’s top-of-the-line, super-expensive speakers, but it is likely to trickle down into mainstream consumer audio systems in the not-too-distant future.
Another promising application of signal processing in consumer audio
is compensating for imperfections in the speaker driver itself. For
decades, manufacturers have worked to improve loudspeakers through
better electromechanical design, more advanced materials, and tighter
manufacturing tolerances. But for applications where costs must be
tightly controlled and where weight and size are highly constrained,
it may be more practical to use relatively poor transducers and
correct their defects with signal processing. Imagine being able to
coax acceptable sound out of the coin-sized speakers in your laptop or
symphonic sound from a $100 boom-box. In the coming years,
inexpensive signal-processing chips will enable these and many other
breakthroughs in consumer audio.
Linley Group Publishes “A Guide to Wireless LAN Chip Sets”“A Guide to Wireless LAN Chip Sets” provides qualitative technology analysis, comparisons, and conclusions on chip sets for 802.11. This new report from The Linley Group—specialists in analysis of networking chips—covers more than a dozen companies, from market leaders to new entrants.
For details, go to http://www.linleygroup.com/Reports/wlanguide.html.
Custom Benchmark ReportsIn addition to the processor analyses contained in BDTI’s Buyer’s Guide to DSP Processors and Inside reports, BDTI has analyzed the DSP performance of numerous other commercially available processors and cores. BDTI can use its vast database of processor performance results to create a custom report comparing the processors of interest to you.
To learn more about BDTI’s custom reports, contact Jeremy Giddings at
giddings@BDTI.com.
BDTI Workshops at the 2003 Communications Design ConferenceWondering which processors are best for mobile multimedia applications? Need to know whether you should use an FPGA or a DSP in your communication system? Join BDTI for two workshops at CDC in San Jose, September 30 to October 2, and find out. BDTI’s CDC 2003 workshops are:
Admission to the conference program of this year’s CDC is offered free
to qualified applicants. For more information go to
http://www.BDTI.com/bdti_whatsnew.html#cdc
BDTI’s “Sounding Board” ServiceThe Sounding Board session was quite valuable for us. We plan to make it a regular part of our product announcement process. — Product manager, semiconductor manufacturer. Don’t just take our word for it. BDTI’s “Sounding Board” service helps vendors of DSP-related products—chips, cores, tools, and software—develop accurate and compelling marketing presentations.
To find out more or to arrange a “Sounding Board” session, contact
Jeremy Giddings at giddings@BDTI.com.
About BDTIBDTI is an independent source for DSP technology analysis and optimized DSP 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 DSP 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.
The next issue of BDTI’s DSP Insider is coming in August. Previous issues of BDTI’s DSP Insider are archived on BDTI’s Web site. Follow the link from http://www.BDTI.com/dspinsider.htm. If you have comments, suggestions, or other feedback about the DSP Insider, please send email to dspinsider@BDTI.com. BDTI’s DSP Insider is a free monthly electronic newsletter published by Berkeley Design Technology, Inc. If our newsletter was forwarded to you and you would like to receive it regularly, please register at http://www.BDTI.com/dspinsider.htm.
If you no longer wish to receive the DSP Insider, send an email
message to dspinsider@BDTI.com with the words “Remove me” in the
subject line.
BDTI’s DSP Insider © 2003 Berkeley Design Technology, Inc. |