Home
About
Help
Feedback
|
The Harrow Technology Report
Insight,
analysis, and commentary on the
Copyright © 2001-2005, Jeffrey R. Harrow. All rights
reserved.
YOUR 'Next Killer Apps!
Listen to this Issue.
Quote of the Week.
CPU Update Update.
PC 'System' Bottlenecks
Abound.
YOUR "Killer Apps..."
On 'Smart Shopping.'
Toy Planes? Not! 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://www.theharrowgroup.com/articles/20021118/20021118.mp3 .
Quote of the Week."All major scientific breakthroughs were scoffed at when they first debuted. To move forward, a scientist has to explore the seemingly impossible."
Marc Millis CPU Update Update.If you read last issue's "CPU
Update" article carefully
(http://www.theharrowgroup.com/articles/ In the
past nine years, CPU clock speed had risen from .066
gigahertz to 3.1 gigahertz, or a 47-times increase
during that time. However, the article I was
discussing from PCWorld.com,
(http://www.pcworld.com/news/article/ This might lead one to question if the vaunted Moore's Law might in fact be slowing down, which would be contrary to everything we've been hearing! I wrote to Martyn Williams, who graciously set things straight for us: "The prediction was made by Intel at the IDF Japan conference here recently. You're right, on the face of it it looks like Moore's Law is breaking down. However what I understood from Intel is that this is NOT the case. Gelsinger [Intel's Chief Technology Officer] said they could make faster processors easily, but they would consume vast amounts of power and generate large levels of heat. Thus, going forward it seems to be that a more precise balancing of issues, like heat and power [... and] processing speed [is required] -- thus the apparent slowing down as compared to Moore's Law. Also, things like the Radio Free Intel initiative may not increase the processing speed but they [may] add to the complexity of the processors. At IDF Japan, numerous Intel executives and employees referenced Moore's Law to me, so I certainly think it is not something they are abandoning or reconsidering. It still seems to be driving the company forward." In a further Email (slightly rearranged here), Martyn added: "I
asked him about the jump being predicted for mobile
device chips and why it seemed smaller than what we
have seen for desktops and [this] was his response
-- basically, it's not just speed anymore.
Gelsinger was making the same point when he said:
"So today, if I was going to look at a StrongArm
core or XScale core [special-purpose low-power
CPUs], could I create a 2GHz or 3GHz XScale today?
Absolutely. It's an interesting change and the big challenge will be for Intel to re-educate users who have got so conditioned to thinking about MHz, or GHz as it is now." Thanks Martyn. This could well be interesting. Some time ago AMD stopped touting the raw speed of its chips because they claim that other aspects of their architecture provide the same end-user horsepower as Intel's "faster" clock speed chips. That claim that has been supported by some past benchmarks, and it has just been supported again by a new round of tests conducted by PCWorld.com (http://www.pcworld.com/news/article/0,aid,106898,tk,dn111402X,00.asp). In tests between very highly-configured 3.06 gigahertz Pentium 4 systems and similar AMD Athlon XP 2800+ systems (which actually run at 2.25 gigahertz), PCWorld.com found that: "Intel says [the new] hyperthreading lets the P4 act as two processors, simultaneously tackling multiple applications or a single application with multiple threads. That should boost performance, but exclusive PC World tests show mixed results at best." ... "The Athlon XP system scored 130 on PC WorldBench 4, besting all three P4 systems as well as the dual-Xeon unit. In fact, the score for each P4 system was either the same or slightly worse with hyperthreading turned on; the Falcon experienced the biggest drop, scoring 127 with the feature disabled and 121 with it enabled." But there's more to this than a few mano-a-mano comparisons; do check out that article for a better understanding. Perhaps Intel may decide to begin going down this road as well, eschewing the "possible" raw speed for better functionality gained in different ways. Won't their Marketing departments have fun...?
PC 'System' Bottlenecks Abound.Following the thoughts we just discussed above, reader Jez Wain suggests: "Though the recent increases in CPU clock frequencies have been spectacular, it has to be said that the 'end-user experience' has not speeded up by anything like the same amount. As you point out, some of this is due to software bloat, with each new release containing more widgets that consume some of our new cpu cycles. However, this same bloat has a secondary effect. As the binary executable becomes bigger, the efficiency of the cache hierarchy diminishes - the cache size is some percentage of the application binary image size, so if the image size increases, the percentage of [available] cache size decreases proportionally. The difference in cache sizes on "consumer" chips such as those produced by AMD and Intel is smaller by almost an order of magnitude than the caches on "enterprise" cpus such as PA-RISC, PowerPC, SPARC or Alpha. For high-power computing, HP (with ex DEC) and IBM understand the importance of cache efficiency. This situation is aggrevated since the relative increases in memory performance and system bus clock frequencies have not [kept up with] that of the CPU. With a memory bus at 200MHz, an L2 cache miss will result in the 3GHz CPU stalling for some 60 - 100 cycles (a memory read uses about 4 bus cycles). The result is that today's consumer cpus wait very quickly. This said, I am impressed by the advances in bus technology. As you point out in your article, just nine years ago, Intel (like all other chip manufacturers) could get a signal across a few millimetres of silicon at 66MHz, and the 200MHz chip is even more recent. Today we send signals over several centimetres of copper wire, in an electrically noisy environment at the same frequency - this is a real achievement and I take my hats off to the engineers who managed to make it work. You also raise a question, towards the end of the same piece, as to what will we do with this computing power. I have been asking myself the same question for some time, though more from a commercial computing perspective than a 'home computing' one. Your neat ideas for how to make shopping less of a hassel would certainly require a fairly sizeable computing infrastructure. But even this system would be small change to today's computers. I am led to believe that the biggest commercial computing application is the New York Stock Exchange which generates some 500,000 transactions per minute. Today you can buy a single SMP system, off the shelf, which handles this load. It reminds me of Marvin in the hitch-hikers guide to the galaxy: 'Here I am, brain the size of a planet, and they ask me to take you down to the bridge. ' In general today's computers work fast enough for today's problems, and there are no new major killer apps on the horizon that will make the system vendors happy. Java, is/was perhaps a shot at this. Being interpreted, Java makes the underlying CPU do more work per (Java) instruction. If enterprise java makes the grade (especially in the face of .NET) then maybe we will be able to use some of those cycles. In the meantime, what we need now is faster memory, faster busses and faster disks. It's only the marketing department that needs faster CPUs." YOUR "Killer Apps..."When I asked you for YOUR ideas
of the next "Killer Apps" -- new applications that
might consume all of the computing power on the
horizon for the year 2010, and so regenerate the PC
business, I had little idea what a deluge of
creativity I would unleash
(http://www.theharrowgroup.com/ Of course this is a Good Thing, and I've been busily collecting and reading and collating the results to share back with you (and hence with the software developers who might 'make them so'.) Overall, these are excellent technologists' views of how to use more of the Moore that we are getting, which I've roughly placed into categories. As you'll see, they cover a lot of ground both in breadth and in depth, and so they offer some serious food for thought for those trying to 'fill up' year-2010, 15 gigahertz, billion-transistor chips.
You've sent me FAR too many ideas to be reproduced here (40 consolidated pages of ideas at last count), but you can click on any category or "idea" link in the structured Table of Contents below, which will take you directly to the appropriate place within the Special Report document. Or, you can go directly to the beginning of the Special Report at
http://www.theharrowgroup.com/
Note that this issue of The Harrow Technology Report continues following the list, below.
Hyperlinked
Table of Contents to:
There's much truth to the old saying that "Software Sells Hardware." Most people don't give a ... er, care about what's under the hood of their shiny new PC. In fact, they don't care much about the PC itself at all. What they DO care about is cost, reliability, ease-of-use, and, mostly, "functionality." They aren't buying (or upgrading) their PCs because of the PC, but because of "what it can do for them, specifically." Give them new "must have" programs that offer easy-to-use, reliable, affordable, and "delightful," dramatic competitive advantages (in their business or personal lives), and the hardware sales will almost surely follow. These are YOUR visions towards that end! Let's 'make them so'!
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
On 'Smart Shopping.'Commenting on our recent
discussion about the intersection of shopping and
technology (http://www.theharrowgroup.com/articles/ "In regards to your piece on smart shopping cart and grocery store technology "Big Brother Update" I thought I would relay information on the same subject that Fred Langa recently published in his newsletter (http://www.langa.com/). The real big brother scenario plays out when the technology can identify the difference between a customer that always buys a certain product and one that may be enticed to do so only occasionally. With the capability already being tested to charge each customer an individual price for an item based on his/her proclivity to purchase it, it will not be long until the customer who always buys a product will almost certainly be paying a higher price than the occasional buyer whom will be induced to buy the product with on the fly discounts. This possibility does not sound so far-fetched. [Consider], as reported by Fred, one of the early grocery store chains that has adopted this technology. [They] have [now] raised all of their meat department prices across the board by 35% -- only to discount the price back to their original 'normal' selling price only for those customers using their shopping card! Essentially anyone that fails to provide the store with extensive marketing and personal data on themselves, gets penalized by paying inflated prices. As you mentioned in your article this technology has great potential benefits to make life a lot more efficient and user friendly; it also has a tremendous potential for abuse." One low-tech countermeasure against this comes to us from reader Paul Hook:
And from reader David Schachter: "According to a Wall Street Journal article from the mid-90's, grocery stores were expected to make more profit from data sales, than from groceries, by the mid-2000's! That's not such a surprising claim when you realize that profit margins for groceries are only 1-2%. However, your suggestion [about having the store's computer optimize my route to increase my shopping efficiency] is inconsistent with how supermarkets are designed: YOUR time efficiency is about the lowest thing on their priority list!
Products are carefully placed to maximize sales. For
example, the dairy department is in the back and the
vegetables are on the sides so that you have to walk
past the high-profit items to get to them. Some
supermarkets have radio tracking devices in shopping
carts to see where consumers go, or surveillance
cameras, or other ways of tracking shopping
behavior, all with the purpose of enticing you to
buy things that aren't on your shopping
list." Toy Planes? Not!Finally, it seems that the small "glow plug" powered model planes that I once flew around the neighborhood so many years ago, have grown up. According to a UPI story brought to our attention by reader Steve Keith, and to an article in Popular Science (http://www.popsci.com/popsci/science/article/0,12543,188271-3,00.html), a Marine may one day squat down on a battlefield, pull five small parts out of her backpack, and assemble them into a 4.3 pound "Unmanned Aerial Vehicle" (UAV). She'll then fire-up the battery-powered engine and toss it into the air. As the little plane flies away at about 45 MPH under her control, she'll watch streaming video of the battlefield on her monitor! This personal "eye in the sky" is very quiet; it's almost undetectable from 300 feet away, and it's small enough that it can be mistaken for a flying bird. In fact, test units have gained the unwanted attention of hawks, which may have been checking out a perspective mate. This UAV is called "Dragon Eye," and it will cost about $3,500 each. The soldier-carried control station will cost less than $10,000. And we thought that "mobile computing" was all about stock quotes and Email and text messaging...
About "The Harrow Technology Report"
"The Harrow Technology Report" explores the innovations and trends of many contemporary and emerging technologies, and then draws some less than obvious connections between them, to help us each survive and prosper in the Knowledge Age. "The Harrow Technology Report" is brought to you by Jeffrey R. Harrow, Principal of The Harrow Group. http://www.TheHarrowGroup.com . Where To Find "The Harrow Technology Report:"
Copyright (c) 2001-2005, Jeffrey R. Harrow. All
rights reserved.
|
