Slashdot Mirror

The "northbridge" of the Intel E7500 supports two PCI-X busses (more information about the chipset can be found here.

The ServerWorks GC series support for PCI-X start from 2 independent busses (the GC-SL) up to six PCI-X busses (the GC-HE). Specs on the ServerWorks stuff is located here.

I'm not completely sure if the AMD Hammer chipsets will include PCI-X support initially, but if one were to give up AGP 8x (which isn't really needed on a server) then you can turn that into a PCI-X bus to support a single 10 Gig Ethernet controller.

Of course, there is still the bottleneck of the memory subsystem which can make or break a high-end system.

We're going to need something to replace PCI before we can use 10Gbit ethernet fully though.

Good point. We need something like 3GIO. Plus something has to be done about the bandwidth between the northbridge and the southbride. Right now it is at 266 MB/sec with plans to increase to 533 MB/sec.

It should be obvious that to burry copper is completely obsolete.

Wrong! Copper is already strung around every city and home in America (probably a hefty portion of the world). And, there's a standard for gigabit over copper:

Deployment Guide
[PDF]

It's limited to 100 meters, but for communities, home networks and any switched network, I don't see a point in passing up what is already laid in the building. For future digs, they could go either way, and I'll agree fiber is the way to go. But let's not ditch copper just yet...it seems to have some usefulness left in it.

So, I do A LOT of USB hardware/software stuff in my disseration work, and I think the Cypress series is by far the best. If you get an elcheapo chip to convert the USB signals then you have to program the protocol manually which is a serious pain (that was the first thing I tried). I gave up and switched to a Cypress EZ-USB because it's the most popular, meaning it has the most support out there. You don't have to use the microcontroller because you can access the USB directly using a DMA to the USB memory location. The microcontroller runs at 24MHz, while the full speed mode (USB 1.1) runs at 12Mbits/s, but the project you describe doesn't need to be faster than 24MHz. For a slow speed application you can just route the data from the USB memory to an output port, and capture the input data on another port. It's a 52 pin package (surface mount) - two 8-bit ports and one 2 bit port on this package.

There's a book released by Intel called "USB Design by Example" which has the EXACT project you're working on. It details all the hardware setup and provides software examples. Its software is in Microsoft Visual Basic and C++. If you want to reproduce that project this is the best way to go. Let the microcontroller deal with the USB protocol to keep it simple.

gtstapler

Careful, the P5 may not like your 6L6 tubes and instead go for some EL34's.... Although they'll probably use 12ax7's

The P5s I have don't care much one way or the other. :-)

("P5" refers to a somewhat older processor than you think.)

A key one is HDCP which makes CSS look like child's play. This extension to DVI will make everyone's life more difficult with devices they use every day-- if Intel's dream of HDCP + DVI being the standard consumer digital video interconnect takes hold.

Two years ago -
JUMPtec AMD 486 DIMM PC 66MHz running a Slackware with apache and pppd, getty etc. and ccam (connectix camera capture software) INSIDE a Omnipoint Redhawk 2000 GSM modem hooked up to a Connectix Quickcam bought on eBay for $40. Linked up to mobilewebcam.com (I let it drop so someone else has got it now). Unfortunately, most of this hardware is end of lifed but it was good while it lasted.
The whole lot spent time between the dashboard of my car and being nailed to a tree looking at my house being built.
The aim was to build a totally autonomous web cam capable of working anywhere. To avoid slashdotting (even a few hits maxed out the connection) we had it uploading JPEGs to a server but it could be accessed directly, no problemo. The only issue we had was the 9.6k circuit switched connection to the ISP would drop every so often, indeed there were telephone numbers that wouldn't even accept 9.6k connections! Nowadays, we'd use a miniscule GPRS module like this one to get a 33k uplink always-on Internet connection.
One other problem was that the picture took quite a long time to scan - it was a parallel port connection. So if you moved the camera, like when I was driving, the picture washed out to white fairly quickly. When stationary it worked fine though.

It should already run on this. Montavista had Linux running on the XScale architecture in February (story here). Also, the XScale CPU can run code written for the StrongArm SA-1110, which powers the HP iPaq, Sharp Zaurus, & other PDAs (specs here).

So does the e740 - looks like these features are part of the XScaleCPU & support chipset.

Jon.

Good point, The bios vendors themselves are not the best providers of documentation... ( AMI for one ).. However some computer manufacturers do document changes:

For example:

IBM
http://www-1.ibm.com/support/manager.wss?rs=0&rt=0 &org=psg&doc=MIGR-41214

COMPAQ
http://www.compaq.com/support/files/server/us/down load/14220.html

HP
http://h20004.www2.hp.com/keeper_rnotes/bsdmatrix/ matrix50451.html

INTEL
http://www.intel.com/design/motherbd/ad/ad_bios.ht m
ftp://download.intel.com/design/motherbd/ad/P04-00 10d.pdf

As far as the bridge offer goes, I'll pass.

Good point, The bios vendors themselves are not the best providers of documentation... ( AMI for one ).. However some computer manufacturers do document changes:

For example:

IBM
http://www-1.ibm.com/support/manager.wss?rs=0&rt=0 &org=psg&doc=MIGR-41214

COMPAQ
http://www.compaq.com/support/files/server/us/down load/14220.html

HP
http://h20004.www2.hp.com/keeper_rnotes/bsdmatrix/ matrix50451.html

INTEL
http://www.intel.com/design/motherbd/ad/ad_bios.ht m
ftp://download.intel.com/design/motherbd/ad/P04-00 10d.pdf

As far as the bridge offer goes, I'll pass.

You shouldn't need to replace your case and power supply

Actually, that's not true. With the P4, Intel now requires an ATX12V PSU, instead of just a "regular" ATX PSU. The main difference is the +12VDC connector on the newer PSU.

http://www.intel.com/home/tech/components/power_su pply.htm

Your questioning of Intel's math abilities intrigued me, so I looked into the new pricing.

As clearly stated on the new pricing table here, the P4M 1.5GHz dropped 26%, the P4M 1.7Ghz dropped 53%, and other P4Ms dropped between those two percentages. Clearly they were stating the range of percentages of the price drops.

Since you might indeed be math impared yourself, I will show you how they got the 53%:

(Orig. Price - New Price) / New Price * 100 = % Decrease
($508-$241)/$508*100=53%

I hope this cleared up the issue. I don't know which is worse, one who spouts off without looking at the facts or one who just bashes a company to get karma. ;-)

The article says that Intel is attributing the price cuts to higher yields, which in turn are due to large investments in its foundries. I'm a little puzzled by this, since this is suggesting that mass-market chip cost actually has something to do with supply, whereas I'd generally assumed that most chip prices were determined by some combination of development cost and demand (i.e., you'll have enough chips; just charge as much as the market will bear and if development is expensive enough you won't have enough competition to bring the price down). The latter is almost certainly true for many server chips. How much is the price of high end mass-market chips actually determined by supply limitations these days?

I would love a case that looks like a piece of hi-fi, something i could place in my component stack (as a media/coms device) and it wouldnt look out of place,my dvd,amp,cd,video,console etc are not in big ugly boxes so why should my computer be any different ?

The FutureClient by Signum Data looks interesting. Cool looks (IMO), heatpipe cooling, near-silent. Now if the price came down a little...

The current PC hardware platform is in dire need of redesign, and I'm not even talking about "features" like the infamous A20 gate. Most cases on the market are badly designed knockoffs with bad EMI and sound insulation and lousy looks. Yes, this goes even for all the expensive aluminum cases I've seen.

Why does the inside of a typical PC look like the site of an explosion in a wire factory?

Why do we still have those stupid power connectors that can be a royal pain to pry off instead of a unified power bus?

Where are those cool slide-in media bays which were part of the PC99 (IIRC) design?

Why do I have to stock sets of metric and imperial screws? (BTW, Torx rules, forget Phillips!)

Does a typical PC really need upwards of 6 fans of varying diameter, RPM and noise emission in order to keep its parts from frying? An intelligent air-duct design would keep down costs and noise emissions.

Wouldn't it be better to have a standard connector for front-panel switches and LEDs instead of the tangle of wires and unkeyed connectors which I always manage to get working only on the second try?

Why do my hands look like I tried to wrestle a porcupine after a motherboard and drive switch?

Intel and Microsoft claim leadership on hardware and software design of the PC platform, and they're both screwing it up badly. I won't even start to bitch about Microsoft, but hey Intel, what happened to those futuristic case designs? (warning - site needs Flash)

-Don

====

This library is mainly aimed at real time computer vision. Some example areas would be Human-Computer Interaction (HCI); Object Identification, Segmentation and Recognition; Face Recognition; Gesture Recognition; Motion Tracking, Ego Motion, Motion Understanding; Structure From Motion (SFM); and Mobile Robotics.

Library Areas:
The areas covered by this library are
Chapter Contents
Image functions: Creation, allocation, destruction of images. Fast pixel access macros.
Data Structures: Static types and dynamic storage. Contour Processing: Finding, displaying, manipulation, and simplification of image contours.
Geometry: Line and ellipse fitting. Convex hull. Contour analysis.
Features: 1st & 2nd Image Derivatives. Lines: Canny, Hough. Corners: Finding, tracking.
Image Statistics: In region of interest: Count, Mean, STD, Min, Max, Norm, Moments, Hu Moments.
Image Pyramids: Power of 2. Color/texture segmentation.
Morphology: Erode, dilate, open, close. Gradient, top-hat, black-hat.
Background Differencing: Accumulate images and squared images. Running averages.
Distance Transform: Distance Transform Thresholding Binary, inverse binary, truncated, to zero, to zero inverse.
Flood Fill: 4 and 8 connected
Camera Calibration: Intrinsic and extrinsic, Rodrigues, un-distortion, Finding checkerboard calibration pattern
View Morphing: 8 point algorithm, Epipolar alignment of images
Motion Templates: Overlaying silhouettes: motion history image, gradient and weighted global motion.
CAMSHIFT: Mean shift algorithm and variant
Active Contours: Snakes
Optical Flow: HS, L-K, BM and L-K in pyramid.
Estimators: Kalman and Condensation.
POSIT: 6DOF model based estimate from 1 2D view.
Histogram (recognition): Manipulation, comparison, backprojection. Earth Mover's Distance (EMD).
Gesture Recognition: Stereo based: Finding hand, hand mask. Image homography, bounding box.
Matrix: Matrix Math: SVD, inverse, cross-product, Mahalanobis, eigen values and vectors. Perspective projection.
Eigen Objects: Calc Cov Matrix, Calc Eigen objects, decomp. coeffs. Decomposition and projection.
embedded HMMs: Create, destroy, observation vectors, DCT, Viterbi Segmentation, training and test.
Drawing Primatives: Line, rectangle, circle, ellipse, polygon. Text on images.
System Functions: Load optimized code. Get processor info.
Utility: Abs difference. Template matching. Pixel order<->Plane order. Convert Scale. Sampling lines. Bi-linear interpolation. ArcTan, sqrt, inv-sqrt, reciprocal. CartToPolar, Exp, Log. Random numbs. Set image. K-Means.

Intel® Image Processing Library (included in OpenCV WinOS download):
Image creation and access (same image header used for both libraries).
Image arithmetic and logic operations.
Image filtering.
Linear image transformation.
Image morphology.
Color space conversion.
Image histogram and thresholding.
Geometric transformation (zoom-decimate, rotate, mirror, shear, warp, perspective transform, affine transform).
Image moments.

Demo Overview (apps that come with the library)
Matlab Camera Calibration Toolbox tutorial
Automatic camera calibration filter
Color tracker/face tracker
Condensation filter fracker
Face recognition using embedded HMMs
Kalman filter tracker
Lucas-Kanade optical flow in an image pyramid

User Contributed Utilities

Windows* Specific
How to find any Direct Show* camera driver with the CAMSHIFT demo
Matrox Meteor* Direct Show capture filter

Linux* Specific
C Code, Non-Specific
BMP* to IPL file reader/writer
Finding the mean and covariance of data sets on disk

====

your post is interesting; because this claims that the p3m burns &lt 0.5W in the 300mhz "battery optimized mode". Even after throwing in the 1W your pc stats page claims for the north bridge chip, the p3m comes in under the crusoe. this claims about 1W under full load (audio/modem/etc). Interestingly enough though, the 440mx is a single component chipset, so now we have to figure in comsumption for a south bridge for the crusoe.

also, the first paper claims &lt 1.0W for the p3m at 533mhz, which is about equivalent to a 677mhz crusoe

anyway, its hard to judge what the total wattage each system would drain on average, and thus how much heat they would emit, but intel is much more competitive than you would lead us to believe.

your post is interesting; because this claims that the p3m burns &lt 0.5W in the 300mhz "battery optimized mode". Even after throwing in the 1W your pc stats page claims for the north bridge chip, the p3m comes in under the crusoe. this claims about 1W under full load (audio/modem/etc). Interestingly enough though, the 440mx is a single component chipset, so now we have to figure in comsumption for a south bridge for the crusoe.

also, the first paper claims &lt 1.0W for the p3m at 533mhz, which is about equivalent to a 677mhz crusoe

anyway, its hard to judge what the total wattage each system would drain on average, and thus how much heat they would emit, but intel is much more competitive than you would lead us to believe.

United Devices along with Intel have designed a clustering program where they use PCs from finding a cure to cancer to modeling the climate..

Only thing that sucks is that they don't have a linux distribution available.. yet

What about ATA-133? SATA will support 150MB/s, will draw less power, use smaller cables, be directly connected therefore doing away with the master/slave and C/S problems. Plus it will scale to 300MB/s in a few years. Read all about it here and here and here.

Intel says the the same

"Based on Intel® NetBurst(TM) micro-architecture, the Pentium® 4 processor was expressly designed to power the media-rich applications that Microsoft Windows® XP allows. At the same time, Windows XP was specifically coded to provide built-in software support for the Pentium 4 processor. Together, Windows XP and the Pentium 4 processor deliver some of the most advanced technologies ever implemented in a mainstream PC platform.

• Intel x86 P3 - 29W
  
• Motorola PowerPC G4 MPC7455 (p15) - 30W
  
• AMD Athlon Model 4 - 55.1W
  

A cpu heat sink HOW heavy? May I suggest you get a better computer next time?

My spare cpu cycles and bandwidth are being used to cure cancer, which I think is a slightly better use of it than for some dipshit's piggyback trojan.

Making links in random words in your post does not help readers find information any quicker when there's no discernable pattern in what they're for

Huh? What exactly did Compaq do with DEC's stuff that was so great? I mean, they're not a chip designer, what on earth are they going to do with the Alpha?

Well, some chip company thought highly enough of the designs of that chip to have purchased the team that was in the process of designing the next generation of Alphas (EV8). The Alpha chip was a legacy of Digital, and continues to be the speed demon for single and multi-chip systems.

The only large PC manufacturer I'd feel comfortable buying from is Dell.

Why? How is Dell superior to Compaq, HP, IBM, or any other PC packager?

That brings back memories!

The 34010 kicked butt! It was used by Atari's Hard-driving game. It had a lot of neat features, including hardware X/Y addressing (i.e. move x,y,pixel), bit-level addressing (you could twiddle any bit in memory, or write a word/byte on any boundry), and built-in simple graphics operations (copy a block of memory, xor source & destination, use larger of the two, subtract, union, difference, add but don't overflow, etc)

But what was *REALLY* cool was the math coprocessor, the 34020. It was blazingly fast (almost, but not quite as fast as the industry-crushing i860 IIRC), but it featured a programmable microcode so you could create your own instructions and get every ounce of performance out of the machine. I'm still looking for a processor that will allow that... we're getting those with modern NPUs (cradle, intel IXP1200), but these generally lack floating point functionality.

OpenCV is Intel's Open Source Computer Vison Library. It's a decent-sized library that helps out with image processing tasks. It's for image processing though - you still have to work out how to capture the video yourself. For that, use V4L (see other comments). One of the included example apps ('CAMSHIFT') includes code to do video capture, but only works under DirectShow on Windows.

If you want to do stuff like extract image features or track moving objects in the video stream, it'll be perfect.

The code is available from the 'opencvlibrary' SourceForge project under a BSD-style license.

Note that the above are for the older, 3.x version of ArcView. Information on the newer ArcView 8 can be found in book and online course formats.

I hope that helps some. What I would ask in return is that when your company gets going with GIS you help promote the use of GIS in Schools (book) by helping any local schools that are interested in getting started with this. School/Business partnerships benefit everybody, especially when it provides the schools with alternatives and additions to the standard Microsoft/Intel curriculum.

tried to download a 20MB file at 0.8kbps

I start twitching every time I see/hear someone mention the concept of "broadcasting" content on the Web -- because it implies that the Web works like high-tech radio. It doesn't work that way, and it never will.

If you want to store objects by location in n-space (like a sparse matrix), you could use a set with the coordinates as the key. A dense storage matrix can be trivially implemented with vector classes, but you cannot to matrix operations on them.

I recently had problems setting up an Intel Digital Movie Creator. The Intel tech support guy actually did research and called me back -- 3 times -- with more suggestions. Eventually, we got it going (I'm going to fly mine on top of a model rocket) and he called back AGAIN to get more system info for their infobase. Total time was about a week; we're talking about a support guy who would actually set up a phone appointment 3 days hence, and keep it. More than once.

This on a discontinued product with a $30 street price.

There's also ATI. I haven't had to call them in a while, but every time I have called the phone is answered by a tech who immediately gets to work on the case. No phone jail, not even an operator to put you in a queue for the next available tech. From dialing to fixed in 5 minutes. Only downside, and it's minor IMO, is that it's a non-free call to Canada.

Considering how Seti has more people than it can handle and last I heard people are doing redudant blocks not just for error checking but just to keep the clients busy.

The intel/united devices program doesn't have as many participants and you get the added bonus of joining AMDzone's team (ranked #3) just to show Intel what you think of their pricing/products.

Click.

Sadly there's no Windows version of the unsupported offer.

Sending material to space is not cheap, so Criswell has studied lunar rocks collected during the Apollo mission years and determined that 90 percent of the aluminum, silicon and glass needed to build solar power plants can be found on the moon.

Let me get this straight. He wants to build a semiconductor manufacturing plant like the ones Intel owns.......but on the moon......that cost $2 billion.......

No wonder why this guy has been pitching this for 20 years without a bite.

Let's see, you can buy 21.5% efficient dual-junction GaInP/GaAs solar cells for $10 each or $17.44 per watt. $2 billion will buy ~115 megawatts worth of power. Save some money for infrastructure and overhead, so 100 megawatts worth of power. To obtain 100 megawatts of power, one would have to buy 174.4 million solar cells which will cover 465,067 square meters. I don't have the link to the NREL site handy, but let's assume that you can get 15 kilowatt-hours per square meter per day which yields 6,976 megawatt-hours of energy per day. At $0.07 per kilowatt-hour, that's $488,320 worth of energy per day.

Now, how many homes do you think you can power off of 6,976 megawatt-hours or $488,320 worth of energy in a single day? Now this guy claims an initial project of $15 billion and then another $135 billion to make it break even. My scenario is for $2 billion. Take that by a factor of 75 and you get 523,200 megawatt-hours worth of energy per day or $36 million worth of energy. Let's subtract the lack of pointing 7,500 megawatts worth of microwave radiation (the same as ~11 million 700 watt microwave ovens on simultaneously) at the Earth and do a risk analysis. Yup, this guy is a lunatic.

Windows 2000 drivers are only available from intel through the mail, with 1-2 weeks delivery. Does anyone know where else I could get these drivers? thx.

This is the correct link. Intel has support thru 2003. They stopped making them as of 3/29/2002.

The IntelPlay site has a good QX3 FAQ as well as the product's 3/29/02 obituary. We can only hope they are clearing out stock before introducing the new improved version, but I doubt it - Intel is discontinuing ALL Intelplay toys, not just the microscope. With MTV style advertising like this it's no wonder it failed to find a continuing market niche as a classic toy - it takes longer than a 10 second attention span to do science. Some gross-out photos are here, as well as a comparison of a QX3 vs.Zeiss dissection scope as well as a comparison of the QX3 and another "inexpensive educational toy" called the Pocketscope. The main Pocketscope site talks about how to add video and lighting to their superior optics. Tinkering with, adjusting, modifying and using the QX3 is discussed here, here and here. More places to buy a QX3 before they go universally out of stock are GlobalMart, Erwincomputers, and Amazon.

The IntelPlay site has a good QX3 FAQ as well as the product's 3/29/02 obituary. We can only hope they are clearing out stock before introducing the new improved version, but I doubt it - Intel is discontinuing ALL Intelplay toys, not just the microscope. With MTV style advertising like this it's no wonder it failed to find a continuing market niche as a classic toy - it takes longer than a 10 second attention span to do science. Some gross-out photos are here, as well as a comparison of a QX3 vs.Zeiss dissection scope as well as a comparison of the QX3 and another "inexpensive educational toy" called the Pocketscope. The main Pocketscope site talks about how to add video and lighting to their superior optics. Tinkering with, adjusting, modifying and using the QX3 is discussed here, here and here. More places to buy a QX3 before they go universally out of stock are GlobalMart, Erwincomputers, and Amazon.

The IntelPlay site has a good QX3 FAQ as well as the product's 3/29/02 obituary. We can only hope they are clearing out stock before introducing the new improved version, but I doubt it - Intel is discontinuing ALL Intelplay toys, not just the microscope. With MTV style advertising like this it's no wonder it failed to find a continuing market niche as a classic toy - it takes longer than a 10 second attention span to do science. Some gross-out photos are here, as well as a comparison of a QX3 vs.Zeiss dissection scope as well as a comparison of the QX3 and another "inexpensive educational toy" called the Pocketscope. The main Pocketscope site talks about how to add video and lighting to their superior optics. Tinkering with, adjusting, modifying and using the QX3 is discussed here, here and here. More places to buy a QX3 before they go universally out of stock are GlobalMart, Erwincomputers, and Amazon.

Q: Do you have Linux drivers for the microscope?
A: The microscope is not supported on Linux* platforms. Thus, Intel does not offer software or drivers for Linux.

Totally agree.

But If you keep your computer on, you are wasting electricity anyway.
Personally, I think SETI is a waste of time.
I would rather use the spare CPU cycles on humanitarian efforts like Medical Research

The really interesting part is this: (quoted from intel.com)
Unified 2MB or 4MB on-cartridge L3 cache. Runs at full processor frequency and is organized as 4-way set-associative with 64-byte cache line size. Fully pipelined and optimized to provide fast access to data at a bandwidth of 12.8GB/sec using a 128-bit wide cache bus.
Now that's cool.

There must be something wrong with the graphs for the e1000 packet size vs. throughput plot, I believe the axis are reversed.-

Also Intel acknowledges that their e1000 adapter have driver issues under linux. This text is from: ftp://aiedownload.intel.com/df-support/2897/ENG/re adme.txt

Known Issues
============
Driver Hangs Under Heavy Traffic Loads


Intel is aware that previously released e1000 drivers may hang under very
specific types of heavy traffic loads. This version includes a workaround
that resets the adapter automatically if a hang condition is detected. This
workaround ensures network traffic flow is not affected when a hang occurs.

This is for the driver verion 4.1.7, released 3/23/2002 (ie. quite new). Older versions had even bigger problems. This might explain why the Intel adapter does so bad in this test. I wish that Intel gets a clue and releases all card specs and GPLs the existing driver so that a true (stable) open source driver could be written and included in the linux kernel. I think the hardware is OK, but the drivers sucks.

"Faraday Cage"

You're catching buzzwords and missing the point. The P-IV processor is packaged in materials not known for their radiation absorption. While the heat spreader is nickel-coated copper, the substrate itself is "Fiber-reinforced resin." (P-IV Datasheet, Page 55)

Plastic.

I have never seen a Pentium (MMX | Pro | II | III | IV) use a grounded heatsink, either.

If you were harboring any illusions that Intel puts shielding in its' processors, please check them at the door, thankyouverymuch. That's what the computer case is for.

If you've ever looked at a Class B (that's home use!) shielded case, you'll see the (unused) external drive bays covered with metal. IBM used to put a very nice braided wire rope gasket on the joints of the PS/1 (among others). You'll also find similar leakage prevention in many rack-mount servers.

Heck, the PS/1 was in the original Pentium days, when processors were running at 200 MHz -- that means a 1.5-meter (nearly 59-inch!) wavelength! All that shielding effort wasn't just for fun, you know.

And, since I'm bothering to respond to all this, I might as well make a point about Faraday Cages:

• Look at your microwave oven. Specifically, the shielding for the window -- or find one that
• has a window if yours doesn't. The holes are quite small. (The energy that oven puts out is at nearly the same frequency as this version of the P-IV, incidentally.)

  Now, what if I were to cut a 3-inch hole in that window? It's easily smaller than the 5-inch / 12.5-cm wavelength. By your logic, no radiation will escape. Would you be willing to turn it on and stand directly in front of it for an extended length of time?

  (Hint: not a good idea.)

If you decide to write the application yourself, and you are using an Intel architecture, then you might consider using the Intel Image Processing Library.

The library provides a set of low-level image manipulation functions in DLL and static form. A part of the API deals with tiling of big images, so that only the viewable fraction of the image is loaded in the memory. The library comes with a demo app that demonstrates its capabilities including the tiling of images.

Second, at 2.4 GHz a signal doesn't follow a wire (or a circuit board trace) like it does at 60 Hz. At 2.4 GHz a wire is more of a 'suggestion' than a 'command'. This is why (radar | microwave ovens | certain satellite communication systems) use waveguides instead of wires. It's also one of the reasons everything isn't running at the same clock speed.

Third, one of the Ten Commandments of /. -- Thou shalt query Google.

• Arcowave AWL-1100P
• D-Link DWL-520
• Intel PRO/Wireless 2011B LAN PCI Adapter
• LG Wireless PCI Card
• Linksys WMP11
• Proxim Harmony 802.11b PCI Card
• Samsung SWL-2000P or SWL-2100P

PCI Cards are installed with the PCB facing in the general direction of the processor (in the ATX spec). I don't know the shielding capabilities of circuit board material, but it sure isn't a solid conductor -- and... many of your traces are exposed to the radiation inside the case. This is where I expect problems and performance degradation to have their roots.

Perhaps you remember a few years ago when it was trendy to install shielding around your audio card for a greater Signal/Noise Ratio? I saw people use copper flashing (the stuff you use to keep your roof from leaking) to construct a box, doing a very nice soldering job, use stand-offs for installation... all to remove a little static. The whole trick was to construct a Faraday cage that would allow the ISA connector (remember those?) as little clearance as possible, without actually shorting it.

We may see a resurgence of that technique.

Intel's P-IV (including 2.4 GHz) datasheet states the power consumption at 49.8 amps @ 1.5 volts. That means nearly 75 watts!

Couple that amount of power with 478 waveguide-like pins to direct it, and you've got yourself a nice little white-noise broadcasting station. Just for kicks, I'd like to see the performance of an 802.11b PCI card trying to coexist with one of these!

How long before some clueless induhvidual brings one of these (in a case with a window mod, thus defeating the Faraday-cage effect) to a LAN party? I give it a couple weeks.

Let's see... P-IV @ 75 watts, vs. 802.11b @ about 1 watt? Which one do you think will win?

The noise floor for 802.11b is going up a few steps, that's for damn sure.

Here is the link to write them and let them know what you think. I myself haven't decided between the "YOU ALL SUCK, FUCK ALL Y'ALL" and the "Dear Gentlemen, I am disappointed to hear..." approaches. But do as you see fit.

Send email to complain.