LISTEN To This Issue.
            Give your eyes a rest, listening at your own rate.

Quote of the Week.
            Computers everywhere!

Forgetting Our Memory.
            We won't even remember what memory is...

The New Age Paper Trail.
            Amazing ways to recover what once was (assumedly) destroyed.

CPU Wars.
            They just keep getting better, faster, and cheaper.

The Full Meal Deal.
            PCs are about far more than just their CPUs.

About "The Harrow Technology Report"

LISTEN To This Issue.

Do you prefer to let your ears do the work of keeping you in-touch with, and thinking about where technology is taking us?  If so, "The Harrow Technology Report" is also available in an audio-on-demand, Web-based, MP3 version. 

If you have an MP3 player on your system (and most do, such as Window's Media Player, RealPlayer, etc.), clicking on the link below will either stream the file to you, or, depending on how your system is configured, it might download the file before playing it.  Alternatively, if you specifically want to download the file, simply right-click on the link, and choose "Save Target As..."

Also, to learn how you can listen at whatever speed is most comfortable to you, check out the FAQ at http://www.theharrowgroup.com/help.htm .

So, if you wish, just click on the following link to listen to this issue!  http://theharrowgroup.com/articles/20020121/20020121.mp3 .

 

Back to Table of Contents

Quote of the Week.

 

"In 2000 alone, 385 million microprocessors were shipped, [but] 6.4 Billion microcontrollers (microprocessors embedded in other devices, such as in elevator controllers) went out factory doors."

Computers ARE in almost everything!

Mercury Research,
quoted in the Nov. 14 CNET News
http://news.cnet.com/news/0-1003-201-7865172-0.html?
tag=dd.ne.dht.nl-sty.0
(This article also contains an interesting read on
the history of the "accidental microprocessor.")

 

Back to Table of Contents

Forgetting Our Memory.

 

That may seem like an oxymoron, but I believe that the day isn't too far away when we will, indeed, forget about the memory we have in our PCs.

Throughout the PC era, memory has always been on our minds, since there was rarely enough of it.  For example, back in the late 1970s when I built my first microcomputer, I could only afford to put in 2 kilobytes of memory (.002 megabytes) because memory cost $62,500 per megabyte!  Even as memory prices dropped precipitously over the years, it always seemed that "enough" memory was just over the rainbow.  Until last month, when reader Bob Pendleton reported seeing "a Fry's ad that showed that I could buy a gigabyte of PC133 SDRAM for less than $100 dollars."  (Memory prices have since been rising -- when I checked Fry's Web site in early January, 1 gigabyte of memory cost $120, and three weeks later it is twice that, or $119 for a half-gigabyte module - http://shop1.outpost.com/product/3140230 .)

Still, even with the recent increases, memory is now so inexpensive that if you can afford a PC, you can probably afford all the memory you need to maximize its performance (adding more memory, to a point, can provide a dramatic increase in performance, especially for Windows XP).  Note, though, that many recent PC motherboards impose limitations on the maximum amount of memory that they will accommodate.  For example, my particular 1 gigahertz PC will accept only a half-gigabyte of memory, since when it was designed (not that long ago), a half-gigabyte of memory seemed pretty outrageous.

My point is, that it's now practical to simply toss in enough memory when we buy or upgrade a PC, and then "forget" about memory, in the same way that today's tens or hundreds of gigabytes of disk space have made the days of constant file pruning (dare I say it) a painful memory.


Semiconductor Skyscrapers.

But if we think that stuffing a gigabyte of memory onto two tiny circuit boards (each about the size of a stick of chewing gum) is something, just wait, as memory goes "3D!"

The January Scientific American (http://www.scientificamerican.com/2002/0102issue/0102lee.html), brought to our attention by reader Andy Mermell, likens the coming changes in memory architecture to the difference between San Francisco and Manhattan.  San Francisco office buildings are generally built close to the ground (due to earthquakes), providing relatively few offices in any city block.  Manhattan, on the other hand, historically took the opposite approach, building "up" to pack a far greater number of offices per block.

Most computer chips have been built using the San Francisco model, basically building transistors one layer deep on the silicon.  (This hasn't kept us from building a LOT of transistors -- Gordon Moore (yes, that one) estimates that we've built over one hundred quadrillion of them so far!)  But this is just the bottom of the iceberg, as scientists and engineers prepare to build future chips more like Manhattan skyscrapers, stacking many transistors one-atop-another.  And this isn't a pie-in-the-sky hope -- companies such as Matrix Semiconductor (http://www.matrixsemi.com/3dtech.shtml?26) plan to have these 3D chips on the market this year!

Just as building a skyscraper requires very different engineering solutions than those for a low-rise building, 3D chips are not a simple extension of today's "single layer of transistors" chips.  The article explains some of the difficulties and some of the innovative solutions that are now changing these semiconductor building rules.  But it's the end result of this work that will be of most interest to you and I:

"Vertical electronics can reduce manufacturing costs 10-fold or more, and the density of 3-D devices should increase at least as fast as Moore's Law, as we add layers."

Matrix' first product is expected to be a single "8-story high" memory chip (a single chip, not a module of several memory chips such as go into today's PCs) that holds 512 megabits of write-once memory, at a cost comparable to magnetic or optical storage.  Matrix believes this will be ideal for "digital film," among other uses.  And if Thompson Multimedia gets its way, we'll get to benefit from this innovation before the end of this year, since the Jan. 8 News.com reports on Thompson's plans to incorporate Matrix's 3D 64 megabyte write-once memory cards, costing $10 each, into digital cameras and MP3 players this year!  (http://news.cnet.com/news/0-1006-200-8404650.
html?tag=dd.ne.dht.nl-hed.0).

And of course this is just the beginning, since Matrix has already proved the feasibility of "12-story high" chips, and expects 16-story chips to follow.

Current 3D chip technologies are not, of course, a panacea -- these chips have innate defects that require "fault tolerant" techniques to work around the expected flaws.  These chips also operate more slowly than conventional "low-rise" chips.  And heat dissipation becomes more of an issue as so many more transistors are packed into the same package.  But this is a very interesting start.

 

Just The Beginning...

Where this will end, of course, we have no idea -- especially since many other companies are working hard to make their own versions of the next huge (tiny) things.  The roster includes companies you would expect, such as IBM and Motorola, but also startups who believe they can "change the rules" big-time, such as Nantero (http://www.nantero.com/tech.html), brought to our attention by reader Sander Olson.  This startup has recently received $6 million in financing to develop carbon nanotube-based non-volatile read/write memory, which they're calling "nanoelectromechanical NRAM."  They believe that NRAM will be at least ten-times more dense than today's typical memory.

The way NRAM works, according to Nantero CEO Greg Schmergel in the Oct. 29, 2001 Mass. High Tech (http://www.masshightech.com/displayarticledetail.
asp?art_ID=51630), is by: 

"...changing the charge placed on a latticework of crossed nanotubes. By altering the charges, engineers can cause the tubes to bind together or separate, creating the ones and zeroes that form the basis of computer memory.

The chip stays in the same state until you make another change.  So when you turn the computer off, it doesn’t erase the memory. You can keep all your data in the RAM and it gives your computer an instant boot."

Additional insights into this technology comes from Schmergel's comments in the Oct. 31, 2001 Red Herring (http://www.events.redherring.com/vc/2001/1031/390020439.html):

"[NRAM] will use an electromechanical approach to storing memory instead of electrical charges. The building blocks are remarkable structures called carbon nanotubes. The smallest of these nanotubes are so thin that atoms must pass through in single file. They have tensile strength greater than any fiber, and 60 times greater strength than steel of the same weight.

In current circuit design, electrical charges are turned on and off to represent ones and zeros in a computer's binary language. Nantero cofounder and chief scientific officer Thomas Rueckes figured out that you could move the nanotubes up and down to represent on and off states. Because the nanotubes are so tiny, "you can store many gigabytes of information on your fingertip."

Do you remember "mechanical relay memory," such as this 1-bit relay-driven binary adder from 1940?

It seems we might be headed back to the days of mechanical memory -- but with a very "tiny" twist!

Overall, Nantero believes that their NRAM holds the potential to:

"... replace all existing forms of memory, such as DRAM, SRAM and flash memory, with a high-density nonvolatile RAM – 'universal memory.'" (http://www.nantero.com/pdf/Nantero%
20Press%20Release%20v6.pdf)

Initially though, according to Schmergel in the Oct. 29 InformationWeek.com (http://www.informationweek.com/story/IWK20011026S0036), their first goal is "merely,"

"... to create a commercial prototype with 1 Gbyte of storage.  But he [also] believes that Nantero will be able to make a 1 terabyte-capacity chip within three to five years!"

Rather a significant challenge.  But if they (or someone else) prove successful, wouldn't that change a LOT of rules...

 

We ALWAYS Find A Way!

From all of these examples and more, one outstanding thing remains clear to me -- that each time we approach fundamental limits or boundaries, creative people continue to find innovative new ways around or through them. 

Thus has it always been, and thus, I believe, it will remain.

Don't Blink!

Back to Table of Contents

Your Feedback is Important!

 

I'd like to understand your interest in The Harrow Technology Report, how you make use of it, and the value you feel it provides to you, your career, and to your company.

Please send your comments to me at  Jeff@TheHarrowGroup.com  .

I look forward to hearing from you!

 

And, if you know of other folks who might find value in "The Harrow Technology Report," I'd appreciate your letting them know that they can subscribe at http://www.theharrowgroup.com/signup.asp .

Jeff Harrow

 

Back to Table of Contents

The New Age Paper Trail.

 

The Enron scandal is revving up, and the hunt for the digital detritus that may "confirm or deny" is moving into high gear.  Which is a good reminder for us, in this day of networked everything and "Email, Email, Email" -- that copies of any messages we send may well reside not only on our own system and on that of the recipient, but the messages may also be laying hidden in the logs and backups of any number of intermediate Email servers, owned by any number of companies, located in any number of countries, that participated in sending a message from here to there.

In a similar vein, most of us know that hitting that "Delete" key (for an Email message or a file) almost never actually deletes it at all -- "deleting" typically just breaks links in the directory structure, marking the old data sectors as available for re-use, but NOT really destroying the data on those sectors.  Forensic computer scientists, commercial firms, and even off-the-shelf software can do a credible job of stalking the seamy back alleys of a hard disk to piece things back together and reveal all.

On the other side of this game of technological escalatio, there are numerous software utilities that do their best to assure that "delete" really does mean "delete."  Instead of just breaking the directory links to a file's (or message's) bits and bytes, these utilities actually write over the old data, sometimes as many as seven times using various patterns that, one atop another, are supposed to render the contents unrecoverable.  The thing is, even these techniques might leave traces of the original bits behind, say on the fringes of the magnetic track on the disk, which might still be readable by the right experts using the right equipment.

In fact, this digital detective work can go even farther, as described in the Jan. 14 New York Times (http://www.nytimes.com/2002/01/14/technology/ebusiness/
14DELE.html?todaysheadlines),

"It is possible to take a disk apart and use an electron microscope to read information from the individual magnetic spots on the surface of a disk that may have been intentionally erased."

So -- in this age of digital business, remember that regardless of the seemingly ethereal nature of the ones and zeros of today's correspondence and records, a "digital paper trail" can be far more persistent than the physical paper trails of old.  Where once, shredding a piece of paper likely ended the trail, in the digital world, a bit may come back to bite you when you least expect it!

 

Back to Table of Contents

CPU Wars.

 

Perhaps, this should instead be titled "The Beat Goes On," since it seems inevitable that our CPUs get faster and far less expensive (per compute cycle) on a regular basis.  For example, AMD has just announced (http://www.amd.com/us-en/Corporate/VirtualPressRoom/
0,,51_104_543~13577,00.html) their Athlon XP 2000+ which runs at 1.67 gigahertz and sells for $339 in quantity. 

The "2000+" is AMD's way of claiming that their chip's architecture enables it to deliver real world performance, for many applications, that is on a par with, or surpasses, competitors' chips that run at or over 2,000 megahertz (or 2 gigahertz).

Which brings us to Intel's announcement, implied above, of its 2.2 gigahertz Pentium 4 (http://www.intel.com/products/desktop/processors/desktop/
pentium4/index.htm?iid=Homepage+Feature_Text), selling for $562 in quantity (http://news.cnet.com/news/0-1003-200-8365659.html?
tag=dd.ne.dht.nl-hed.0).

It's a good thing that this competition exists, but it's also confusing -- if you're looking for a new high-end system, how do you choose which chip is right for you? 

One way is to examine benchmarks, such as those recently run by ZDNet Germany (http://www.zdnet.com/anchordesk/stories/story/
0,10738,2836825,00.html), which shows the Athlon ahead for many mainstream business applications, but the Pentium taking the lead for compute-intensive tasks such as gaming and video editing.  Your mileage, of course, will vary.

Most of us don't need these fastest chips for our daily computing activities (aside from serious gamers, and people doing compute-intensive tasks where the speed increase could directly translate to time saved and higher profitability.)  But I expect this to change once some enterprising individual spawns the next "must have killer app" that drinks CPU cycles like Kool-Aid.

The day will come, I promise, when the thought of a mere 2 gigahertz CPU will seem ridiculous.  Remember that just four years ago, $2,500 Pentium 2 systems running at 266 megahertz with 32 megabytes of memory, were common.  Yet try running today's operating systems and applications on one...

Back to Table of Contents

The Full Meal Deal.

 

Finally, speaking of increasing computing power, I admit that we're getting rather used to the massive number crunching capabilities of our commodity general purpose CPUs, such as Intel's Pentium 4, AMD's Athlon series, and the PowerPC.  But contemporary PCs don't rely exclusively on that "general purpose CPU" for all of their calculations -- PCs also often contain task-specific processors to enhance highly compute-intensive functions such as generating sound and graphics.  And it's fascinating to realize just HOW MUCH computing power resides in these "ancillary" chips.

For example, Steve Jobs recently announced that future Macs will offer Nvidia's new GeForce 3 video chip as an option (http://www.fileplanet.com/index.asp?file=56348), and this video chip alone contains 57 million transistors (more than the 42 million transistors in a Pentium 4)! 

Fifteen years ago, when Pixar introduced its then-amazing computer-generated video titled "Luxo Jr." (http://www.pixar.com/theater/shorts/ljr/short_320.html), it required 75 hours of Cray supercomputer time to render EACH SECOND of photo realistic animated video.  Today's GeForce 3 chip can now render similar video in real time! 

How does it pull this off?  Unlike the (mere) two billion operations per second of our current mainstream CPUs, the GeForce 3 Ti 500 video chip runs at just under ONE TRILLION specialized operations per second!  (http://www.nvidia.com/view.asp?PAGE=geforce3)

Those numbers are all well and good, but what does that mean to those of us willing to shell out the bucks ($600 extra on the Macs) for this bleeding-edge graphics capability?  Nvidia offers several movies of the impressive results, which you can view on any contemporary system, at http://www.nvidia.com/view.asp?PAGE=pg_20010529517600 .

Bottom line?  When it comes to "computing power," don't just focus on the CPU "main course" -- the side dishes, ranging from video to audio to modems and to other chips, really round out our "computing" meal.  And the additional computing power that gets added to this menu every year assures that our palates have many, many tasty surprises yet in store!

About "The Harrow Technology Report"