Slashdot Mirror

The most intersting small-form-factor board I've seen recently is the new SensorTile from ST. It provides a 32bit ARM processor with FPU, bluetooth and microphone in a cm2 headset-scale pcb, with various docking boards for USB, microSD etc development.

How many free compilers are there for stm8? Just sdcc :-( so we chose to pay $5000 for iar which im not happy about.

I would start here

Although to be fair, my first question is: why ST? I have looked at them before, and every time I look, I can get a 32 bit uC from Cypress with more IO, and pay less unit cost than the cost of the 8 bit ST chips.

http://www.st.com/web/catalog/... Cost: 10 Euro. That gives you an ARM Cortex with plenty of resources, Arduino compatible headers, debugger on board, and the tools are free and you can use http://www.mbed.org/ with it. And yes, the pins are labeled.

http://www.st.com/web/catalog/... Cost: 10 Euro. That gives you an ARM Cortex with plenty of resources, Arduino compatible headers, debugger on board, and the tools are free and you can use http://mbed.org/ with it.

ST makes several ARM M4F based dev/eval boards with built-in JTAG and a few additional chips thrown in to play with (I think accelerometers and MEMS microphones are common). They cost around $15-$20... go to http://www.st.com/web/catalog/... and check the box for STM32F4 under Supported Devices.
So, with what I suspect is the benefit of manufacturer loss-leader subsidies on the Discovery boards, why would I spend $40-$60 more on a dev board?

You can build one yourself. Assume the BeBox geek port:

One "GeekPort" (37-pin D-shell)

        An experimental-electronic-development oriented port, backed by three fuses on the mainboard.
        Digital and analog I/O and DC power connector, 37-pin connector on the ISA bus.
        Two independent, bidirectional 8-bit ports
        Four A/D pins routing to a 12-bit A/D converter
        Four D/A pins connected to an independent 8-bit D/A converter
        Two signal ground reference pins
        Eleven power and ground pins:
                Two at +5 V, one at +12 V, one at -12 V, seven ground pins.

My current favourite are STM32 discovery boards, i.e.:
http://www.st.com/web/catalog/tools/FM116/SC959/SS1532/PF254044

(costs $10)

It even has a User USB port, you can use. Then you have to invest some time implementing the whole
communication chain to your PC. ST generally has a standard peripheral driver library that can help you
with all the stuff you can do with the chip in question.

Apart from the powersupply stuff the bebox offers, you will find everything else and more in the microcontroller.
If you want to build an external geekbox you would want to add a separate power supply. Then you could use the microcontrollers PWM outputs
to regulate some voltages for a power supply output, or use some dc/dc converters for fixed voltages.

From the top of my head I wouldn't know what to do with it though, but I do need a power supply unit, some measurement means, and some stimulation means for circuits.

Why are you posting a link about the raspberry pi, to prove that the STM32 microcontroller has HDMI?

Try reading the datasheet for the thing actually being talked about: http://www.st.com/internet/evalboard/product/252419.jsp

There is NO HDMI. Not to mention it only has 1mb of flash and 192kb of ram.
This device isn't even in the same class of computing devices! It's a micro-controller, not a micro-computer. Micro-controllers do not natively do video out at all, you would need one fast enough to manually bit-bang composite video signals out of an IO pin. You would be hard pressed to find a micro-controller fast enough to bit-bang HDMI signals, let alone output it.

Pointing out that a completely different and unrelated device such as the rasp-pi has HDMI, does not at all follow that every microcontroller and 8 pin chip out there can also output HDMI.

I'm wondering if the model A will really have much of a market.

The end of the market that the A might have been useful in may well have been overtaken by the top-end of the M-series ARM processors, especially with companies like STM now pitching boards like the Discovery STM32F4 for $20 or so.

Yes, it's got less RAM, less MIPS and so forth -- but it *is* 100% open and incredibly capable for what it is.

I know everyone loves Arduino, but I don't get it. If you think you can explain it to me, first read specs of Raspberry Pi($35 and runs Linux, has Ethernet, USB, etc.) and STM32F4DISCOVERY ($15, 210 DMIPS, FPU, 1MB of flash, 192kB of SRAM, has USB host/device/otg, onboard 3-axis accelerometer, mic, stereo DAC with speaker driver, JTAG debugger also built in).

With those two on the market, I don't see what Arduino is for...

Bus Pirate: good for looking at communication waveforms to debug problems. ($35)

Logic Sniffer: For more complex problems than the above, allows looking at parallel signals.($50)

Raspberry Pi: Tiny ARM11 700MHz CPU with powerful graphics, 10/100 ethernet, USB2.0 host (2 ports), HDMI out, and GPIO connector. Boots from SD card. ($35)

MSP430 Launchpad: inexpensive microcontroller development platform ($4.30)

STM32F4Discovery: Development platform for powerful microcontroller. ARM Cortex M4 with FPU, 168MHz (210DMIPS), Ethernet MAC, 2xUSB host/device/OTG, etc. etc. Board has stereo audio DAC with speaker driver, USB Micro-AB connector, 3-axis accelerometer, digital mic, 4 user LEDs, two pushbuttons (one is reset), and onboard debugger which is supported by open source tools. ($15) <--- take that, arduino

This is old stuff so high/low will only be 0V and 5V.

Generalizations are always false. For example:

The M27W401 has been designed to be fully compatible with the M27C4001 and has the same electronic signature. As a result the M27W401 can be programmed as the M27C4001 on the same programming equipment applying 12.75V on VPP and 6.25V on VCC by the use of the same PRESTO II algorithm. [...] Programming with Presto II consists of applying a sequence of 100us program pulses to each byte until a correct verify occurs

Even some modern parts (Atmel AVR MCUs) support high voltage (12V) programming. This link explains what it is for, and the datasheet has far more to say.

Compared to Arduino's Atmel (2,32kB) the Maple STM32/ARM (20, 120kB) has way more RAM and flash (also far more than the TI MSP430 series, while not requiring that much more power). The Maple RET6 gives 64kB SRAM /512kB flash and DACs rather than just PWM for only $10 more than the regular version. A version with FPGA on-board is in development. The Maple Mini which is just 2.02 x 0.72 inches and emulates a 40-pin DIP for breadboarding should be shipping in a few days.

Maples are capable of driving QVGA LCDs, doing modest signal processing (see LeafLabs wiki for a guitar pedal project), controlling motors, doing data-logging and complex control and much more.

While they can be pressed into service for doing traditional computing, microcontrollers are distinguished mainly by their built-in peripherals such as timers, ADC, DAC,sensors, encoders, communications etc. The STM / Cortex devices do very well on this measure, particularly the higher-end parts. The Maple forums are active and helpful; some ongoing topics are on using real-time OSes and interfacing with SD card flash memory.

For a cheaper start to working with the STM ARM chips Mouser carries a rather nice-looking, hackableboard with USB, LCD, blinkenlights, buttons, and capacitive touch sensor slider for just $15.

Lots of interesting stuff out there in the world of micro-controllers, and now lot of it get available at reasonable prices. Not only as those dreaded $999 development kits.

If you look for something more powerful the STM32VLDISCOVERY http://www.st.com/internet/evalboard/product/250863.jsp, is a nice alternative at about $10. You get a modern and powerful ARM Coretex M3 with 128 KB Flash and 8 KB RAM. With lots of nice peripherals included.

Download developer documentation, it's all in there. Accelerometer - it is LIS331DL chip, first hit in google, ffs.

They'll just start adding them together.
UltraUltra low power
UltraSuperMicroMini low power
PicoPicoPicoPicoPower

Or we could skip all that and do what ST does; Embellish a bit and call it "zeropower" (which is trademarked no less).

Zeropower NVRAM - Which of course is battery backed, and uses... power.

Stories like this frequently conflate the smart card goings-on with the system functions.

In this case, the newsy bit about the smart card is they apparently have a new protocol for authenticating from the smart card. For those that don't know, there are many kinds of smart cards including ones that have an operating system on-board. Their protocol is probably employed on top of the smart card OS. Yes, you too can write your own authentication protocol and use it on a smart card.

The backend system appears to have new automagical features related to the status of the employee. Don't confuse the two like the summary has.

OT, I have always thought that "the way forward" in infosec was loosely decentralized smart card infrastructure, but the powerful among us like their power optimized and centralized. Too bad two, the only smart card developers left work exclusively for gov't contractors.
Even further OT: A 'fun' OSS project for those inclined would be to port a BSD to one of these low-cost suckers. http://www.st.com/stonline/stappl/productcatalog/app?path=/comp/stcom/PcStComRPNTableView.onClickFromProductTree&primaryheader=Smartcard%20ICs&secondaryheader=ST32%2032-bit%20Smartcard%20ICs%20for%20Mobile&subclassheader=ST32%2C%2032-bit%20Flash%20Microcontrollers&subclassid=1192.0&count=3&producttype=

In theory, these have a crypto accelerator: http://www.st.com/stonline/stappl/productcatalog/app?path=/comp/stcom/PcStComRPNTableView.onClickFromProductTree&primaryheader=Smartcard%20ICs&secondaryheader=ST19%20Smartcard%20ICs&subclassheader=ST19%2C%20Crypto-Processor%20Solutions&subclassid=1118.0&count=4&producttype=

Stories like this frequently conflate the smart card goings-on with the system functions.

In this case, the newsy bit about the smart card is they apparently have a new protocol for authenticating from the smart card. For those that don't know, there are many kinds of smart cards including ones that have an operating system on-board. Their protocol is probably employed on top of the smart card OS. Yes, you too can write your own authentication protocol and use it on a smart card.

The backend system appears to have new automagical features related to the status of the employee. Don't confuse the two like the summary has.

OT, I have always thought that "the way forward" in infosec was loosely decentralized smart card infrastructure, but the powerful among us like their power optimized and centralized. Too bad two, the only smart card developers left work exclusively for gov't contractors.
Even further OT: A 'fun' OSS project for those inclined would be to port a BSD to one of these low-cost suckers. http://www.st.com/stonline/stappl/productcatalog/app?path=/comp/stcom/PcStComRPNTableView.onClickFromProductTree&primaryheader=Smartcard%20ICs&secondaryheader=ST32%2032-bit%20Smartcard%20ICs%20for%20Mobile&subclassheader=ST32%2C%2032-bit%20Flash%20Microcontrollers&subclassid=1192.0&count=3&producttype=

In theory, these have a crypto accelerator: http://www.st.com/stonline/stappl/productcatalog/app?path=/comp/stcom/PcStComRPNTableView.onClickFromProductTree&primaryheader=Smartcard%20ICs&secondaryheader=ST19%20Smartcard%20ICs&subclassheader=ST19%2C%20Crypto-Processor%20Solutions&subclassid=1118.0&count=4&producttype=

Sure, the boxes decode an "HD" signal. Yes they scale the HD signal down to 480i. You are right about this.

You are wrong about future pricing of these boxes - history is on my side here.

Here is a press release for a Microtune MT2131 chip that integrates analog NTSC, DTV, and digital cable reception capability onto a single chip:

http://www.broadcastingcable.com/article/CA6311888.html

The chip's cost: $2.40 per chip (and this is from 2006 - they are probably cheaper now). No "magic pixie dust" needed.

Here is an HDTV decoder chip from 2004 that cost $18 back then:

http://www.st.com/stonline/press/news/year2004/p1494p.htm

This article details entire system on a chip designs that fell to $15 at the end of 2007.

http://findarticles.com/p/articles/mi_m0EIN/is_/ai_n25455222

Don't you think chips like this will enable set top converters for less than $40?

Technology history is full of examples of expensive stuff becoming really cheap, really fast. Why would DTV set top boxes be any different? You'd be a fool to believe otherwise.

Do you honestly believe that prices will go lower than $40 if the government is giving away that amount of money for each box? No businessman with a brain in his skull is going to charge less than $40 until the money dries up.

My post was meant to illustrate that this commoditization process can now occur naturally since the artificial prop holding up pricing has now been removed.

-ted

Electrolytic capacitors are one obvious place cost can be cut.

There are a couple of topologies that are applicable. Most computer powersuuplies are (I think) single-switch forward converters. The topologies with more switches provide better performance, but more switches means more expensive transistors.

Higher frequency switching generally provides more stable output, but requires "faster" transistors and transistor drivers, which again, are more expensive than "slower" ones.

I see that you're really excited about iPhone, applications, etc.. Here are some facts though:

All these so called "level" apps use builtin LIS302DL accelerometer/motion sensor, which outputs values in range 0-38h for 0-1g accelerations. Do the math and you'll see that despite what apps says/shows, it cannot give you precision more that about 1.6 degrees tops (pdf spec)

Disclaimer: I am iPhone developer and I like developing for it - but people, please do some reality checks sometimes...

Well put. The theremin, in many ways, is more mysterious than this (by some mysterious definition of "mystery") because it essentially uses the user's body itself as a capacitive element in an RF circuit whereas the iPhone accelerometer is a straightforward (albeit cleverly made) direct mechanical effect. If you don't want a toy, you can get real Moog theremin kits for under $400 or the real thing for under $2k. Nevertheless, I do think this is a clever use of the accelerometer in the iPhone. The real test will be if a composer creates an iPhone Concerto in Em with this new instrument.

I don't feel this has to do anything with Taiwan.

In any case, the chips are not going to be manufactured in China.
These will be manufactured by STMicroelectronics on 65nm process in their Crolles Fab in France.

No thats not true.
Although the design is Chinese, the Godson processors will be manufactured by STMicroelectronics, which is a French-Italian company.
The processors will probably be manufactured in Crolles, France on the ST 65nm process. The backend packaging is done in Singapore and Malaysia using RoHS compliant package design.

Here is brief descriptions of the cpu (PDF). It's chinese-developed 64-bit MIPS, has 2 FPUs, 2 ALUs, 64K/64K L1, 512K L2. And consumes 4W@900MHz. It has a builtin ddr2-667 memory controller, PCI-X bus and no builtin video/USB/etc.

Nice processor, but IMO Nvidia Tegra is more suitable for a netbook; Ars Technica writes: "Tegra ... dissipates less than 300mW during HD playback." And has all peripherials integrated on the chip.

Main problem with this netbook is only 4 hours autonomous work, while 7-9 is much more suitable: I can take it to work without charder, etc.

Yeah, because there haven't been 386, 486, and other systems on a chip and Via doesn't have a 1-watt processor anywhere to be found. This is not the first 1-chip chipset for all of the x86 line. That's bullshit. An SoC is even more integrated than just having the chipset as one chip. Somebody never read the old Computer Shopper before it slimmed down. SoC solutions for x86-compatible systems have been around more than a decade. The summary is bad, because TFA does not say this is a first for the x86 line.

You're right that even low-powered x86 chips like the C7 and the Geode line are generally no match for ARM and XScale. MIPS I'm not as familiar with for power usage purposes. It'd be nice if that question was answered, but I'm afraid it'd be summarized incorrectly too.

2005 article on anx86 SoC
another 2005 article about a different x86 SoC
2004 product page for an already obsolete x86 SoC
Linux Devices list of x86 SoC solutions, some dated to 2000
2000 Register article about the year since Cyrix released an x86 SoC
Chipslist page showing availability of AMD processor with 80188 features plus DMA, watchdog timer, serial ports, and I/O pins in 1995
article on the National Semiconductor Geode (the owners of that line before AMD bought it) thin client system-on-chip

And the best proof of all: an archive of a 1996 story on the AMD Elan,which featured a 386, ISA bus, serial UART, memory controller, power management, and PLL hardware ON ONE CHIP

Read the paper

http://www.st.com/stonline/products/literature/an/ 10122.pdf

and this one

http://newslab.csie.ntu.edu.tw/~johnson/public_fil es/R-FLO436_Chang.pdf

Because I wouldn't mind getting a flash drive. I'm just a bit skeptical.

Well it wouldn't surprise me if some flash disks get this disasterously wrong to be honest. Some of the lifetimes on page 6 of the Chang paper look a bit low for comfort - around 4 years for NFTL.

Power consumption specs for programming PRAM are not stated in many places but
I was able to find a reference here:
http://www.hitachi.com/New/cnews/051213.html

And for comparison to flash memory, here is the 512Mb 1.8v part from ST Micro:
http://www.st.com/stonline/products/literature/ds/ 10058/nand512r3a.pdf

The specs do not line up exactly.
PRAM: 100uA at 1.5V for programming each bit cell
FLASH: 8ma at 1.8V for programing one page (256 bytes), internally rebuffered in SRAM

Right from the source: http://www.st.com/stonline/products/literature/ds/ 11115.pdf These things are not exactly new. They are used in the automotive sector, or for "stabilizers" in camcorders.

A good picture of a two-axis accelerometer can be seen here: http://users.wpi.edu/~cfurlong/me-593Mech.html (second picture down). Sensing is usually performed by capacitive combs, structures which act as capacitors, with their capacitance varying with displacement.

MEMS accelerometers have dropped in price in recent years because there's a big market: the automotive sector. A typical new car needs two accelerometers, one for the traction control system measuring roughly plus-or-minus 2 to 4g, and one for airbag deployment measuring more like 50g.

Two big manufacturers are Analog Devices and ST Microelectronics, though others exist.

The high demand of the automotive sector has driven prices right down; sensors which would have cost hundreds of dollars in the past can now be purchased in bulk for less than $4. In fact, you could order one right now; component retailers will sell you one for less than $15.

A friend of mine did an internship at STMicroelectronics in Italy last year, and he told be about the phase-change memory they were developing. This non-volatile memory is based on special materials which can assume two different stable solid states, one amorphous with high resistance, the other crystalline with low resistance, and can be thermically switched between the two. As far as I understand, you can create a memory cell using a tiny spec of this material and an addressing transistor; then you can read it by applying a low current (there will be a large voltage drop in the amorphous state, and a small one in the crystalline state), or write it by applying a higher current (which heats the cell enough to change its state; the final state is determine by the temperature reached). Since the memorization element can be made very small, high densities can be obtained, and because of its tiny size it takes a very little time to heat and cool, so transitions are fast.

I just found an article from 2004 about STMicro's developments; as far as I can tell, Samsung's new memory is based on the same principle. Actually, the entry on phase-change memory on Wikipedia seems to confirm that it is, and lists other companies that are working on this kind of memory, so expect a large influx of PCM devises as more manufacturers bring their products to market.

That chip you're looking at is a STMicro "SMOOTH" combo driver, most likely slightly customised for the application. This particular chip has two power regulators and a serial interface to the microcontroller (most likely I2Cish) in addition to the motor drive stuff.

Also, brushless DC motor drivers that have the drive transistors and the PID controller in the same package have been around for years (Hitachi and SGS were making them back in the late 1980s/early 1990s); the trick was getting them on the same chip as the coil driver, which is more like a BTL audio amp than a motor driver (Seagate actually did use a car audio amp, the TDA1210 I think, in the early ST4000 series drives back around 1984).

-lee

The grandparent is right. Flash systems intended to be used for a filesystem incorporates a layer, usually called the "Flash Translation Layer", which maps writes to blocks from the OS filesystem layer to actual pages of flash memory.

For that matter, filesystem flash is usually "NAND" flash, which higher bit density (and thus lower price for a given size) than "NOR" flash. One problem with NAND is that it tends to have more manufacturing defects and bit errors. So, NAND devices are usually constructed with extra sectors, just like a disk drive, and the FTL software is responsible for detecting bad sectors and mapping around them.

Here's an app note from a more or less typical flash vendor that talks about it.
http://www.st.com/stonline/products/literature/an/ 10122.pdf

That is a reasonable price. Right now, MRAM replaces battery-backed SRAM: like the ZEROPOWER series from ST Microelectronics, and a 4-Mbit version (M48Z512A) costs $45 in quantity, and the MRAM chip won't take up huge real estate with a gigantic DIP package.

At $25 in quantity for a 4-Mbit chip, it's about a factor of 5 higher than conventional SRAM. I'd guess that a factor of 5 in cost reduction isn't crazy to expect.

Too bad this chip didn't come out say, five to ten years ago - otherwise you likely would've been seeing it in video game cartridges for a while now.

At the moment, 32GB of NAND flash is about $580.

http://www.dramexchange.com/

A 40GB hard disk is about $58. Which is a pretty bad price differential.

But the cost of NAND flash should drop 43% per year over the next 5 years

http://blogs.zdnet.com/ITFacts/index.php?blogthis= 1&p=9615

So that 32GB NAND flash device should be $580 * ( 0.57 ^ 5 ) = $34. That's less than a small hard disk, and you can probably charge a premium based on the low power consumption / small form factor. My guess is that people will pay $100-$200 premium for a NAND flash laptop, so I'd expect to see 32GB NAND flash ultraportables in the next couple of years.

I checked the write rate to the Physical Disk on Windows in perfmon (this counts writes to the block device driver) on my work machine, and it's not too bad, about 100K/sec averaged over an hour of building a massive chunk of software. Mainstream desktop use should have a much lower rate. E.g the laptop I'm writing this on is writing at 3K-20K/sec average at the moment. Most of that is short 40K spikes, with a small duty cycle too, so an daily average should be lower.

This paper has a formula for disk lifetime
http://www.st.com/stonline/products/literature/an/ 10122.pdf

i.e. lifetime in days is

Size of flash * Number of erase cycles * FS overhead
/
Bytes written per day.

Let's assume that the 32GB array is 8*32Gbit chips. Each chip is then 4GByte

Plugging in the figures, we get

4GB * 100,000 * 0.7
/
100KB * 24 * 60 * 60

I get 33981 days, or 93 years. This is mostly because with wear levelling, the lifetime is proportional to NAND flash array size, since wear levelling should spread erases evenly. If you could wear level over the whole 32GB, it would be 744 years!

It's actually pessimistic, since the fs overhead is lower on Inode based filesystem like NTFS or ext2/3. E.g in their example, the file needs 5000 clusters in the FAT, or 20K. On NTFS, file extents are stored as runs - i.e. a contiguous file will be one extent in the 1K Inode. Most of the time, file extents are not growing too.

At the moment, the fastest NAND has a per chip write speed of 10MB/sec, and reads about 100MB/sec. I think you could do some kind of RAID like approach, especially for reads. And since we're talking about a 8*32Gbit chips for this device the read speed should be 800MB/sec, neglecting speed ups from new interfaces and process shrinks. And bad blocks could be tracked on a finer granularity too, so you could continue to use a erase unit until all the blocks went bad. And 1 million erase cycles is probably not impossible to achieve in 5 years, since some NAND devices do that as you say.

People often quote pagefile writes as a reason for NAND being unusable, but I'm not really convinced. Even with a completely unmodified system, a worst case filesystem choice (FAT32 with small clusters), and an application which constantly grows files, and a NAND chip where every block has a 100% fail rate after 100K erases, the lifetime of a large NAND disk is still orders of magnitude greater than a hard disk, which is likely to die in 5-10 years or so at this level of usage, since it wears from both read and write.

They do actually debug their products before releasing them.

Nokia phones are relatively quite bug-free compared to other brands.

(I know this from people who supply them with components.)

I loved my Kyro 2 for the fact that a 50 dollar card at the time gave better peformance than a 60 dollar ATI or Nvidia. SIS, Matrox, ATI and Nvidia; is that really all? What other great graphics companies no longer exist? I can't think of any at the moment.

The recent--and constantly improving--development of write-leveling in the Flash firmware and the OS filesystem drivers has made this essentially a non-issue. By distributing repeated writes to different cells, modern Flash possesses a far greater resiliency than that of Flash produced before wear leveling was introduced. For more information, check out the this whitepaper.
I believe it quotes a lifetime of something like .55 days versus 49 years for repeated writes to same logical block or sector (remember, with wear leveling enabled, the physical cell you're writing to may be different each time).

For filesystem-level wear leveling (i.e., filesystems designed for flash memory), check out YAFFS and JFFS2.

For information about wear-leveling in general, see its Wikipedia entry.

- Roey

This is utter bullshit.

Eh, no, it isn't.

I was looking for some more concrete evidence on power specifically for the CPU and from the datasheet, at 266MHz they quote 1.9W max power. The STPC Elite SoC datasheet (CPU from the example you used) states 4.8W for the x86 at 133MHz.

Yes, it is fucking bullshit.

First of all, your link for STPC Elite doesn't point to a STPC Elite page. Second of all, the STPC Elite does NOT use 4.8W but 2.3Watt max @ 133 MHz

Yes, perhaps that's at half the clock speed, but it also adds PCI, IDE and floating point.

In any case, that is still NOT several orders of magnitude less power.

The whole world treats the US as damage and goes around it.
US treats the ROW(1) as damage and goes around it...
Sounds more true to me.
(1) ROW: Rest Of World

Check your facts.

Inmos
Motorola

This is really just about adding high density flash to an existing smart card platform. Other then having alot of flash this (16 bit CPU, 4-8K RAM) card is just like most other JavaCards out there (such as in your cell phone or AMEX Blue card). The innovative smart cards these days have 32 bit CPUs such as the P9SC648 from Philips and ST22N256 from ST Micro. The Philips card is alot more powerful then IBM/Sharp's card and still has 512 KB Flash. The ST card has 256 KB Flash and 368 ROM and is shipping now for $4 to $5 in quantity.

If you want to learn more about future memory technologies which could replace today's SDRAM and flash-memory try searching for "MRAM", "FeRAM" and "PCRAM". The latter is probably the most promising one due to the (expected) ease of design and manufacturing ("Samsung says that it will enter full-scale production with the technology in 2006").

More information about phase-change (->PCRAM) materials can be found at E*PCOS website.

STMicroelectronics, for one. Another large company you don't generally hear anything about (which is the way they like it).

I've been researching chipsets for digital TV. Here are my links to current hardware products:

STMicroelectronics System on Chip (2) Get Linux here
ATI Xilleon 220 (Products)
Sigma Designs Digital Media Processors (Products)
IBM PowerPC405 STBxx (Zarlink [2], Araneo)
Texas Instruments DM642 DSP (i3 Mood Box , X-Designs Flikit + Softier MediaLinux)
NEC EMMArchitecture2 (Galaxis + LinuxTV , PRISMIQ + Linux)
Equator Technologies BSP-15 boards
Via CN400 (Mini-ITX Board), PM800 and PM880 (w/ HDTV for Pentium 4) , ShowShifter HMN, Soyo Multimedia Ready Motherboard (with TV Tuner, $129.99)
Toshiba TX System RISC (MontaVista Linux)
Windows chipsets:
Intel 815 VisionPlus terrestrial box (Korean OEM)
AMD Geode (CoCom)
ARM (Samsung, etc.)
Digeo X-Stream (Paul Allen company)

Sony actually built the PS1 into the chip that handles the controller ports, not the main CPU(emotion engine). It's more likely that the Xbox2 would have a System on a chip, but Microsoft would still need to emulate all the NVIDIA crud. This would be quite a feat, legally and technically.

The G5@2Ghz may be able to emulate the CPU in software. It looks like Microsoft is at least looking at the route(they bought Connectix but I doubt this is possible). The graphics emulation would still be a bitch.

From what I understand, Sony is almost ready to start fabbing a chip which combines most of the PS2's chips(sound,graphice,etc) and will start using it in PS2s and then in the PS3 to allow backwards compatablilty

Sun's description of Throughput Computing and their approach of putting multiple processor cores reminds me of what Inmos tried to do with the Transputer before they became STMicroelectronics. The idea was to have many small processors positioned close to each other, communicating between each other closely. I seem to recall seeing transputers on eBay a while back for huge amounts of money. By all accounts, a transputer board was a very useful piece of kit for the right appplications!

Great article, but you can improve it for us with just a little bit of html, making links is not hard:

You can find it with many slides at http://online.itp.ucsb.edu/online/colloq/lewis1/

Some of the recent research, and the progress made by startup companies is summarized at:

http://www.konarkatech.com/news_articles-forbes_no v.php

http://www.konarkatech.com/news_articles-solracs-h ybPV.php

http://www.st.com/stonline/press/news/year2003/t13 55h.htm

http://www.nanosolar.com/advantages.htm

... ST Microelectronics already supply devices that mix programmable logic, memory and IO from their Programmable System Device range. But there is something of a reluctance for commercial designs to incorporate them because they're single source components. Why risk being unable to make your product in the future because you've used a specialised component in your system which has gone obsolete - especially when there's a plethora of available direct drop-in replacements for a discrete solution (EG separate programmable logic and memory).

Official company press release.

The ST team is also developing low cost solar cells using a full organic approach, in which a mixture of electron-acceptor and electron-donor organic materials is sandwiched between two electrodes. The nanostructure of this blend is crucial for the cell performance because the electron-donor and electron-acceptor materials have to be in an intimate contact at distances below 10 nm. ST plans to use Fullerene (C60) as the electron-acceptor material and an organic copper compound as the electron-donor.

"These R&D activities, which exploit the expertise we have in nanotechnology, complement and augment the commitment that ST has made to be a CO2-neutral company by 2010," says Coffa. "In addition to ensuring that our own industrial activities have minimal impact on the environment, we are developing many new technologies that we hope will bring substantial ecological benefits."

ST has a website