Slashback: Zoning, Linking, Fooling

Tonight Slashback brings you updates (below) on the video card ATi isn't really putting out, home-brewed electronic multi-room temperature control, NPR's linking policy, and more. Enjoy!

Welcome to the Fantasy Hardware League Regarding our post on the allegedly upcoming Radeon 8500 MAXX, reader eyelove yu writes: "This pic is fake, as many people have suspected. HardOCP.com (on front page) quoted Rubeena Hussein of ATi as saying,'"We have no current intentions of making this or similar boards.'"

Soon we will be able to assemble an entire system created in Photoshop. Yay.

Or you could roll down the windows ... vt@home writes: "As a followup to the earlier story, here is a system that not only allows to monitor the temperature throughout the house and draw nice charts, but also does already have computer controlled vents and even allows to control the A/C unit. Basically, this is a do-it-yourself zoning system, for under $500. Of course, the source is GPLd ;)"

Next week, the sidewalks will practically be free for public use. juanfe writes: "It's not like they really had any power to enforce their previous one, but NPR modified their Terms of Use on June 27. Now, linkers do not have to submit a form asking for permission, but NPR "reserve the right to withdraw permission for any link". More commentary from others.

Nothing like hundreds of angry bloggers threatening to withhold membership contributions to their local station."

Raising a stink to the power of 10. Snarfangel writes "After seeing Yet Another Slashdot Article extolling the virtues of meretricious metrification ("Isn't it Time for Metric Time?"), I decided to fight back the only way I know how -- by subjecting an innocent website to the Slashdot effect: This site goes into great detail about the importance of being Ernst (or at least Max Karl Ernst Ludwig) Planck, especially his system of units that only depend the fundamental constants of the universe -- the speed of light, the gravitational constant, the Planck constant, and the charge of the electron. With appropriate scaling, you get a unified measurement system that is not only more logical than Le Systeme International d'Unites, but is also much better for calculating physics problems in your head.

After all, if we are going to go to all the effort to change our measurement system, why not use that same effort and get the system *right* the first time?"

On a different note, Colin LeMahieu writes "I noticed your post on metric time. I stumbled across this while looking for various computer timing related articles and found it pretty interesting. This might not be as popular as metric time, but it seems to make more sense. The whole system is based on time as a fraction of a day; it even has the scientific measurment on how to re-produce the time, as with any scientific measurement."

Comments

Look at the pic of the Radeon "MAXX"

[Cutriss]

Specifically, look at the screws on the heatsinks of each GPU. They're at exactly the same orientation on both. Someone copied the one on the left, shrunk it a bit for proportion, and copied it onto the card after rearranging the PCB a bit. Notice also the distortion in the upper surface of the heatsink, where it doesn't mesh very well with the voltage regulator behind/above it.

I always wondered about units of measurement...

[gusnz]

...and why this hasn't already happened.

The meter, for instance, was originally defined as one ten-millionth of the distance between the north pole and the south pole. Although now the Earth has been measured more accurately so it's off by a bit, and it's now defined by the length light travels in a vacuum in a very short time.

But really, why are we basing measurements on all these arbitrary values anyway? Like the Imperial system originated from the dimensions of some king's thumb or similar, pretty much every measurement ever devised and in common everyday use is derived from non-universal values, which have no practical upshot -- if we want to measure the Earth, we're going to include some decimal places anyway.

Personally I think this, if adopted, would make scientific calculations a bit easier. It's annoying to have to remember several different conversion constants for gravity, charge, gas constant (8.314 or similar?), and so on. And perhaps without all the continual conversions, relationships between different physical principles might become more readily apparent...?

But I guess the downside is that some calculations are always going to have funny conversion constants, especially in the non-Physics world (Avogadro's number in chemisty perhaps for instance?). So even though the metric system isn't perfect, it's the standard so we might as well use it (although this could be the web developer in me speaking). It would be too much change for too little benefit to rescale the entire number system -- convincing the general populace would be just about impossible, especially considering how much trouble some countries are still having adjusting to the metric system ;).

Re:I always wondered about units of measurement...

[os2fan]

Back in the seventies, I played around with a system where the principle fundemental constants were powers of 10, eg

light speed = 1,000,000,000 ft/s
elect const = 0.000 000 001 'F' / ft
magnt const = 0.000 000 001 'H' / ft
gravitation = 0.000 000 001 lb s^2 / ft^3

Such a system is easy to set up, and produces practical sized units. The nifty thing about this is that one could convert pounds and coulombs with a foot ruler, since the size of the foot, pound, and charge unit directly is in proportion to time. So a mars-ruler laid up against an earth-ruler converts pounds etc. The replacement for Volts, Ohms Watts, and Amperes are not changed from planet to planet. The only trouble is that the thing's hard to set up for practical use.

On the other hand, I did try to look for a 'better' system. I did manage to get eight constants working in a google-system. In essence, the process of dimensional analysis is to let things like L, M, T and I have numeric values, being powers of 10^100. The set I used after much study is L=1E1100, M=1E73300, T=1E100, Q=1E32200. So a kilowatt is 1E75203. One can then work with a wide range of units, eg tonne = E73303 becomes coherent.

You can do the same thing with the fine structure constant, and an assortment of natural constants as well. Instead of powers of 10, you use powers of 137.0359895, or its square root. The relevant units are:

L 1K1100 = 137.036 bohr radii
M 1K73300 = 137.036^2 electron mass
T 1K100 so that c = 1K137.036^3
Q 1K32200 = 137.036 electron charge
t 1 th so that m_e c^2 / k = 137^4

These units refers to one boron-sized molecule at atmospheric pressue, ~ 10 K. Most of the numbers come out as they should: avagadro's number in this system is 10.3 (ie 137.036/1868).

It still does won't be used in science because of the way scientists works. Something like "cgs units" or "atomic units" is of their name.

Fake graphics and dual GPU cards

[yakfacts]

This time, the fake GPU card would have fooled me. There are a couple things that look wrong, but it was a good enough job that I would have been fooled had I not known it was a fake.

There was a fake post here in 2000 where somebody took an Adaptec 2940 card and tweaked it a bit, then claimed it was a Russian-surplus vector-based supercomputer-on-PCI card. Ignoring the fact that the fake graphic was obvious (you could still see the Adaptec logo and QC stickers on the card), I could not believe people would fall for a "cray on a chip" from Russian surplus. While Russia is a fine country with a great history, they are not known for their high-tech electronics. This is the same country that was still uses tube computers and radios in the mid-1990s, and used to buy new pinball machines just so they could pull the 68000 CPUs. If the Russians had any infrastructure to develop such a bleeding-edge device, the certainly would not be selling it. I posted my feelings then and got flamed for it.

But I could fall for the ATI card. ATI has a history of Dual-GPU cards. I strongly disagree with the poster who said "dual is not as good"; depending on how it is done, it can be much better. Don't use Windows NT as your baseline for multiprocessor applications. Design an application (in this case, a driver) that expects to see certain CPUs in certain places and hardware that automagically divides the load. There are good ways to do this if you ALWAYS know what sort of hardware resources you will have. Systems that don't (standard Windoze or Linux applications) will suffer greatly as they try to adapt on-the-fly.

Re:Fake graphics and dual GPU cards

[toybuilder]

Do you have a link/biblio-reference to the info about the Russians buying the pinball machines?

One of my favorite Russian-CS-is-screwed is the story about the metric chips... This Byte article alludes to the original story... In short, the Russians stole western-technology and produced knock-off copies using "the metric inch" -- except when their poor-quality copied failed, they couldn't use real (stolen?) chips to repair their machines.

Re:Vinge's Second-based TIme

[Chelloveck]

I really liked this system too, especially for a space-faring culture which has no need for marking time as integer fractions of the rotation of an arbitrary blue-green planet. Seems to me that it'll make a whole lot of sense to use something like this when we get permanent off-planet colonies. (Especially Martian colonies, where a day is close enough to an Earth day for the residents to live by the Martian light/dark cycle, but just enough off to bollox calendars between there and mother Earth.)

Of course, my favorite part about this system is Vinge's description of when the calendar began...

Second by second, the Qeng Ho counted from the instant that a human had first set foot on Old Earth's moon. But if you looked at it still more closely... the starting instant was actually some hundred million seconds later, the 0-second of one of Humankind's first computer operating systems.

"Beginning of the epoch" indeed!