Well, not exactly. There was a pretty good explanation in another thread about how a modern cellphone chip might theoretically be as powerful as, say, an old Pentium, it can’t actually perform as many operations.
Can’t remember which thread it was now.
The missile thing sounds dubious at best, but the computer “recycling” is quite real. I don’t know about Malaysia, but I know for certain there are whole towns/villages in rural China that are piled 30 feet high with old computer parts. Old computers are shipped in, stripped and sorted by the poor residents of these places for minimal pay, with little to no regard for the safety and health of the environment or the workers.
I know that’s not really what this thread is about, and I’m sure you’re all well aware of the reality of the non-missile part of the OP.
The other thing that suprises me about the comment (which sounds completly spurious to me, but then I know a lot of laws aren’t present in the US)
is this
Is it really legal to collect electronics for recycling and then ship them off-country?
Even if you don’t have enviromental protection laws, and even if there is no law mandating proper recycling, wouldn’t that fall under fraud?
I don’t doubt that some shady companies pocketed the money for proper recylcling and then went the cheap illegal way - that’s what happens every so often in Europe, so we need police to watch out, catch and fine them heavily. But 100%?
The other reason this sounds dubious to me is that currently, with metal prices at such a premium, proper recycling really is worth it to the company. Hereis a German Top-Science show explaining how much you get from electronics:
Out of one ton computer platines the belgian experts can extract 100 grams of Palladium, 250 grams of gold and one kilogram of silver. That doesn’t sound very impressive at first - but with current gold prices, it’s a small fortune. (Translation by me)
I call BS
iirc (so correct me if I’m wrong) the 486DX has the coprocesser onboard. It was the SX that needed the separate coprocessor.
The only likely scenario is some kind of reverse engineered embedded platform in production somewhere that’s missing some discrete components. Think of Iran’s mostly defunct F-14 based air force thats been dying for US parts. F-14 parts are now shredded per DoD request.
Still, thats pretty unlikely. If you can make the embedded board, the supporting chips, and everything else, you can make or source the main CPU. Not to mention it would be a little silly to even produce that today. The leaps in video/signal processing alone would make some 1980s missile platform a pretty lousy by today’s standards. If you want a 1980s level basic missile then there are probably better ways to go about it than reverse engineering a very old design and digging around for home-level non-hardened chips.
I think this kind of thing plays well with the anti-green pro-war Fox News crowd. Voting BS.
$9100 in gold, $1675 in palladium and about $550 in silver at current prices. I imagine collecting a ton of recycled computer components costs about that much anyway.
That’s what I was going to say… why would you bother trying to scrounge up old-ass chips, when anything modern is going to be an order of magnitude faster, and the cost difference would be negligible in the context of a modern missile, even a Chinese one. (what’s a couple hundred bucks when the whole missile costs 10,000 bucks?)
I remember a year or so ago there was a professor building a supercomputer out of video cards. Which is a brilliant idea. They are high efficiency processors perfect for doing certain mathematical calculations they are optimized for. They are also perfectly modular and bussed for getting instructions in, crunching quickly and spitting results out. So all you need to do is custom build a coordinating motherboard*. And For anything that can be easily paralellized(which a lot of supercomputer type stuff can be) you can get to super computer speeds for a couple thousand bucks off the shelf.
*Of course this part is not off the shelf, but many electronic engineers around the world could do it.
I recall reading that computers used in the shuttle and (presumably) the space station need older microprocessor chips. The chips made in the past ten years are too unstable in space because of EMF and radiation.
http://www.cpushack.com/space-craft-cpu.html
cite for chips in space.
I’m thinking that the need is for existing weapon systems, that is, you already have the Patriot system. No one here is building a system from scratch with obsolete parts. That’s not to hard to believe, since the weapon was sold to foreign countries. Remember, the weapon isn’t just the missile, it’s the supporting ground equipment also (radars and targeting). If the system is designed around a certain processor, then just replacing it with a AMD Phenom isn’t so simple, it may have cascading effects that made the weapon useless. Sure, a modern processor is far more capable than the 486, but what does that matter if the system doesn’t work?
No idea if it is true or not, just saying that it is believable. The entire system has to work, in the field under extreme conditions (this isn’t a nice office setting). Sure, you might make it work with a new Core Duo. Can you ensure that it works after being stored at 100 degrees for 4 years, transported for 5 weeks in a salt fog and deployed at 10 degrees below zero? Because the original processor, no matter how primitive by today’s standard, can do just that.
Missed the edit window.
Edit: That being said, it occurs to me that the original contractor (Raytheon? Someone else?) would have support contracts in place making this story dubious.
See? I can’t make up my mind.
I think the SX was physically the same chip, but with the coprocessor disabled (perhaps manufactured on one production line and split into DX and SX in QC).
Isn’t the real danger here not the chips themselves but the contents of the computers hard drive? A lot of people taking computers to be recycled don’t necessarily realise that this could result in someone stripping this out and recovering data.
I could believe that it might be worthwhile in some parts of the world for enterprising criminal types to buy up used hard drives in bulk and data-mine them for useful information before selling them on.
Um, collecting? Are you thinking of the recycling people driving door-to-door with a van to get the stuff?
The comment quoted in the OP talked about recycling, which I assume to mean electronics or computers collected either at the city’s waste resource station (or whatever you call it in AE), which will send it on to the special recycling station; or the sellers of computers who have to take the old stuff back and also send it on.
Because a new law passed here in Germany several years back requires all sellers of electronics to take back old stuff they sold and properly depose of it without danger to the enviroment; since most sellers don’t want to build their own center, they pay a fee to a professional facility. That’s why my quote talks about a Belgian company - they have specialized in the boards (after the first steps are done in Germany), and so are highly efficent in their extracting/ operating costs.
This is correct. The SX and DX were manufactured on the same line. Semiconductors have a horrible rate of failures during the manufacturing process (many more fail than work straight off of the line, believe it or not). If the CPU portion passed but the coprocessor failed, Intel would completely disable the coprocessor (by blasting the coprocessor’s power and bus connections with a laser) and sold it as an SX. If the coprocessor passed all of its tests it was a DX.
The math coprocessor chip in the 486 days was the 487. Similarly, the 8086 coprocessor was the 8087, the 286 coprocessor was the 287, and the 386 coprocessor was the 387. Unlike in previous chips, the 487 was actually a 486DX with a different pinout. When you added the “coprocessor” chip, what it really did was disable the 486SX and ran completely out of the 487. So, in a way, there really wasn’t a dedicated math coprocessor chip in the 486 days like there were for previous generations.
Out of curiosity do modern CPUs still have these high failure/low yield rates for units coming right off the line?
There are two reasons for this.
The first is that spacecraft and the like take a long time to design. So, you start out with the latest and greatest whiz bang processor for your spacecraft’s brain, and your software guys start writing code for it. By the time the hardware guys have built the ship and the software guys have done their thing, that latest and greatest whiz bang processor is now a couple of generations behind the current technology, so the processor is “old” before the spacecraft has even left the ground. Now let’s say you have something like the space shuttle, which has been flying since the 1970s. It just keeps flying and flying and flying and that processor that is in it doesn’t change. Meanwhile, the latest and greatest whiz bang processors in common use have continued to develop. Now, that processor that was just “old” when the shuttle started to fly is “really really old” by modern standards.
You don’t just upgrade the motherboard on a space shuttle. Whatever is in there stays in there. Changing the processor to something more modern requires a huge redesign and test effort.
The second reason you see older processors in space is that they have to be radiation hardened. Modern processors are harder to radiation harden because the logic gates and wire traces are so small. A physically “bigger” transistor is harder to flip into a different state. The same amount of energy can much more easily flip a bit in a more modern and smaller transistor. Figuring out how to radiation harden a processor and testing it all to make sure it works also takes time, which is why space qualified processors lag far behind desktop PC processors.
Oh yeah. They certainly do. Modern processors start with a yield rate well below 50%. That improves a bit over time as they improve their manufacturing processes. Hitting 70% on a mature line is considered really good.
CPU manufacturers are always in sort of a balancing act. Making chips faster means making the transistors and such smaller, which pushes the limits of their manufacturing capability. If they push it too far, they get a horrible yield and even though they may have the better/faster chip, they may not be able to make enough of a profit on it. If they don’t push enough, they have a chip that they get higher yields out of but their competitor ends up with the better/faster chip. It’s a tough game to play.
The NVidia GT300 graphics chip yields in 2009 were under 2%.