» ARM's latest processor cores narrowly beat out Freescale's embedded processors and Toshiba's highest density Flash memory on the technological significance scale but were ranked third on the usability scale.

(1)
ARM
ARM readies cores for control, data processing tasks
eeProductCenter's Marty Gold says: "ARM is broadening its product offerings beyond the well- established 16/32-bit embedded RISC processors with two new types of licensable core-based solutions: one is a multiprocessor core that specifically handles control tasks; the other is responsible for heavy data processing jobs. Targeting the needs of applications executing multiple tasks at the same time such as with consumer entertainment and convergence devices in the home and car, ARM has begun licensing its MPCore synthesizable multiprocessor, which can be configured to contain between one and four processors delivering up to 2600 Dhrystone MIPS of performance. For applications that demand more performance than today's general purpose DSPs, ARM is offering its OptimoDE data engine technology. "
Readers say: » "Nice product, but my applications are too low in volume to interest ARM. Volumes are in 1000's/year. I'll probably use this when someone packages a general part." » "There are more 16-bit and 32-bit problems then are 64-bit problems." » "It took a while for DSPs to become a developer's choice mainly because of tools. ARM offers great tools for its ARM7/9 cores and if the tools are equally good for OptimoDE cores then certainly it would be a cost saving for our physical layer software." » "There will be two factors that will determine how usable this core is for general applications. First is the cache hit ratio, which needs to be high to avoid memory bottlenecks and the second is the ease of coding. Most likely hand coded assembly will have to be used. And, the power consumption is still too high for me to consider this." » "Intriguing." » "I prefer to use off-the-shelf parts. We typically do not have the time available to use core logic as part of a larger design and we do not have the volume to justify it as well" » "Good evolutionary step. Consideration of low operating power with high processing power is wise." » "My experience with ARM cores has been very positive but overall it's difficult for this set of cores to differentiate itself from the sea of offerings out there." » "Multiprocessor cores is defintely the trend." » "Symmetric multiprocessing with Linux is an attractive offering for an SOC device."
USABILITY RANK: 3

(2)
FREESCALE
Freescale expands ColdFire line for industrial control apps
eeProductCenter's Marty Gold says: "The ColdFire embedded processor line, developed by Motorola as the successor to its well-known 68000 family should score some points in market share with the launching of four new product families with more than twenty processors. This is important in that ColdFire is now one of the anchor product lines of Freescale Semiconductor, the new spinoff of Motorola. Targeting such embedded applications as large-scale factory automation, precise medical instrumentation, robotics, security, retail point-of-sale scanners and terminals, among others, the new 32-bit MCF547x, MCF548x, MCF523x and MCF527x families offer system designers a reduction in memory requirements, power dissipation, system board size and costs, according to Freescale. There are several notable features of the launch including the first use of the new V4e ColdFire core. "
Readers say: » "» "Motorola continues to provide highly integrated embedded processors that are well suited for a number of applications." » "A little pricey" » "If tools are good then for client applications replacing a DSP with this can be considered." » "These are slick. I'd like to use them in something, but I think industrial control will go with PLCs or PC104/mini-PCI systems running Windows, because that's what they're used to" » "The MCF5272 was the first part to bring 10/100 Ethernet to the embedded space at a resonable price. The newer 527x and 547x parts are even more useful in new designs." » "Technically this looks like a nice processor, but, personally speaking, there is always a large learning curve to switch over." » "Looks very promising." » "I'm a Motorola fan since the 6800 days... glad they are still 'leading edge'! A bit pricey... " » "Speed, price, features, compatibility. Works for me. " » "The 54xx series stands out due to the MMU and hardware encryption. A 388-pin BGA seems gratuitous to me. The price is high compared to ARM, but the Linux development support might outweigh such concerns..." » "Nice and nifty processor - but out of time. They should have had it 5 years ago. "
USABILITY RANK: 1

(3)
TOSHIBA
90 nm in a Flash
eeProductCenter's Jon Gabay says: "The industry's use and standardization on 90 nm technology is manifesting higher performance and densities across the board. A case in point is Toshiba's newly announced 4 Gigabit NAND Flash parts. It wasn't too long ago that 1 gigabit and then 2 gigabit parts were introduced. Achievable densities have always been possible using stacking technology, and several opted for this solution to keep ahead of the competition. While stacking technology offers higher densities and the possibility to mix different types of memories (such as Flash and SRAM and DRAM), associated costs are higher. Often, package heights are higher as is power consumption since each individual die has to implement I/O drivers, bias circuitry, decode logic and so on. But, if roadmaps are accurate and manufacturers are diligent, monolithic flavors of the technology will prevail and hopefully replace stacked parts where applicable. The other side to the coin is that achievable density increases by stacking the new parts. Both are evident and possible with Toshiba's 4 Gigabit Flash parts."
Readers say: » "I would like to know what is the yield rate for a 90nm process at 300mm wafer and what is the signal integrity." » "Stack packaging has come a long way. I'd like to see more with regard to capacitance and assembly hurdle solutions." » "Are we getting close to an album-on-a-chip? Could these be accessed fast enough to provide two CD quality audio streams in real time?"
USABILITY RANK: 2

(4)
ST
16 and 32 Mbit Devices Added to Serial Flash Family
eeProductCenter's Jon Gabay says: "The benefit of small dense non volatile storage is well known and understood by embedded systems designers. While many microcontrollers and microprocessors now have embedded Flash on chip, the need to go external has not disappeared. Not every micro allows runtime manipulation of Flash. Many require the micro to go into a special program mode by powering up in a special configuration and taking commands and data over a serial port. Even if you do have the potential to rewrite Flash dynamically while running user code, Flash has always been block limited. This means, for the most part, every read or write is a block function, not a true random access function. It becomes a time and code blackhole to go through these hoops every time a single byte for configuration or user preferences is changed, especially when this operation is done on the micro itself. Serial EEPROMs (and Flash) have given designers a small footprint solution to this while keeping costs relatively low and not adding too much complexity to a codesmith's life. Evolution has yielded the next step forward in this technology: STMicroelectronics has announced M25P16 (16Mbit) and M25P32 (32Mbit) devices to its Serial Flash Family that up the bus frequency to 50MHz. This will help to reduce power up initialization times and makes these parts more feasible for cached code storage. Especially as the density has been steadily creeping higher and higher."
Readers say: » "I've already designed this technology into our lastest product and look forward to using even larger devices in the future." » "This is one of those products you want to find a use for." » "The 16Mbit or 32Mbit devices appear to have many advantages over the 4Mbit I am familar with." » "The trend of moving from parallel to serial Flash interface will continue to accelerate because both the ASIC and the Flash benefit from reduced pin count and hence cost. Upon power up the code is shadowed into embedded RAM for high-speed execution. The trade-off of a slower interface is minimal compared to the large cost reduction. » "We use a lower capacity part in one of our products and have hoped for a larger capacity part."
USABILITY RANK: 4

(5)
ANALOG DEVICES
ADI rolls voice-enabled network processor
eeProductCenter's Robert Keenan says: "Analog Devices has a network processor that combines packet, voice, firewall, and security processing in a single package, a chip that ADI says will deliver smaller footprint to multi-service and gateway system designs. There's a clear push in the carrier community for the development of residential gateway systems that support VoIP and data capabilities as well as security technology like firewall. Through the development of the Blackfin Fusiv family, ADI is making is easier for designers to realize these residential gateway design goals. The first device in the Fusiv family, subbed the Fusiv-Vx 200, is developed around ADI's ADSP-2100 digital signal processor core, which is used on chip for processing VoIP streams. "
Readers say: » "This product category has the potential to launch new billion dollar markets worldwide." » "ADI cannot deal with customers that are not in high volume production." » "Would like to see a chip like this include wireless capability, particularly for the price." » "I think I see a potential application in Homeland Security for this type of processor." » "Number of voice channels supported as well as the compression algorithms supported would be nice information to provide" » "Too expensive at the moment - will look at in future to see if pricing is lower."
USABILITY RANK: 7

(6)
TEXAS INSTRUMENTS
TI beefs up C64X DSP line with lower priced devices
eeProductCenter's Marty Gold says: "TI has begun sampling two new devices in its TMS320C64X digital signal processor product line at lower price points than other members of the family. They also pack additional on-chip memory and peripherals functions. The TMS320C6410 runs at 400 MHz with 160 Kbytes of total memory while the 500 MHz TMS320C6413 is a 500 MHz part and embeds 288 Kbytes total memory. Both are packaged in a 23 mm x 23 mm flip-chip BGA package (288 pins, 4 rows, 1.0 mm BGA pitch), which maximizes channel density and allows four layer board routing. "
Readers say: » "» "I've used earlier and costlier predecessors. The price point is very nice." » "TI continues to control the market with excellent DSPs and an industry leading tool chain. Until competitors (such as ADI) are able to match TI's overall offering, they will continue to run in second place." » "I would be very interested in these parts if there are good tools available to migrate away from the TI 54xxx series of DSPs." » "Allows four layer board routing. Thanks. Now, please, make the development software cheaper." » "Great! Another feather in TI's cap, this may obviate the need of high level of optimization, which was a necessity for other expensive C64x DSPs required." » "Sound like great products!" » "I am more interested in even higher speed DSPs. Never can get enough horsepower."
USABILITY RANK: 5

(7)
LSI LOGIC
LSI Logic' s DSP core targets high volume apps
eeProductCenter's Marty Gold says: "LSI Logic's new low-end ZSP200 DSP core is being pitched to system designers as the core that fills the processing gap between hardwired logic and general purpose processors. It's among the growing number of processor solutions that are trying to attack high volume applications such as disk drive controls, industrial controls, toys, electronic games, VoIP terminals and security devices that now require more signal processing capabilities and flexibility than what's been available from either hardwired logic or other processors. Among the processors targeting these and other similar applications are the general purpose DSPs from Texas Instruments, Freescale Semiconductor and Analog Devices, among others. "
Readers say: » "» "We do not have the time or the volume to justify the use of this product"
USABILITY RANK: 6

(8)
PMC-SIERRA
MIPs-based MPU employs fast system interconnect for comm. apps
eeProductCenter's Marty Gold says: "PMC-Sierra's RM9150 microprocessor combines the E9000 64-bit MIPS microprocessor core with the new system interconnect called the Fast Device Bus for embedded control processing applications such as Ethernet routers and switches, line cards, network attached storage, industrial control, high-end laser printers, imaging, and high-end consumer applications that require multiple high-speed interconnects and single-chip integration. The Fast Device Bus interconnect was defined to fully leverage the performance of a 1GHz E9000 processor and support an extensive set of high-throughput peripherals, such as a 200 MHz 64-bit DDR SDRAM controller, a 600 MHz HyperTransport interface, as well as two Gigabit Ethernet MACs, and dual PCI interfaces. "
Readers say: » "This processor is too expensive for use in our application. Our products require the programmable serial interface provided by devices such as the Motorola PowerQuic II series of processors" » "The GDI sounds ok, but I would also wonder how much overhead is there and why wouldn't I go with a hyper transport interface?"
USABILITY RANK: 8

(9)
IDT
Tri Port Devices Bridge and Rate Buffer Between Independent Processors
eeProductCenter's Jon Gabay says: "Odds are, any of us embedded system designers have had to interface to other independent processors along the way. While simple registers which are memory or I/O mapped can be useful for transferring status and/or tokenized commands, it is a slow and cumbersome way to transfer data blocks. This is compounded by data intensive applications which many have multiple processes running simultaneously (or sequentially) each needing to share its own blocks of data. There are solutions. While various multiport devices exist, typically, dual port memory devices are accepted as standard building blocks. IDT has upped the ante by introducing a tri port device family which it hopes will become a standard since so many of today's designs will use multiple processors. "
Readers say: » "Applications and functionalities are growing so this is an ideal product; however, price is scoring more over price-performance." » "This is a good part for medium to low volume applications. I would like to read what the 1K pricing is. 500K pricing is out of most people's volumes and rings similiar to when a car advertisement says 'only one at this price.'" » "I would like to see a low-cost 52-pin, TQFP, dual port device, 2K X 8, $2 in 1K volumes, $1 in high volumes. " » "Ever since John Schmitt proposed development of dual port RAMs at TI Houston in 1977/8, I've been waiting for market acceptance of multi-port devices. Good show."
USABILITY RANK: 9

(10)
STARCORE
DSP core gets mobile media upgrade
eeProductCenter's Patrick Mannion says: "StarCore LLC has equipped the latest version of its digital signal processor architecture with 25 new instructions, a partially interlocking pipeline, memory protection and an extended interrupt scheme. StarCore's V4 architecture will be followed up by the SC2000 core, with a six-stage pipeline and dynamic branch prediction. While wireless is the headliner, the new architecture and core also target converged consumer devices, wireline communications, voice processing and general-purpose DSP applications. With plans afoot to announce its V5 architecture later this year and the corresponding SC3000 core in early 2005, StarCore says it will roll out three architecture families in less than 2-1/2 years. StarCore was spun out 18 months ago from what had been a collaboration between Agere and Motorola. "
Readers say: » "Need to see the available tools and time required to migrate from TI DSPs." » "The connection to Freescale is a plus, but need more technical information. Evaluation kits? Prices?" » "Too big and power hungry."
USABILITY RANK: 10