Dan's Data letters #129Publication date: 30 October 2004.
Last modified 03-Dec-2011.
I was wondering, if you had US$2000 right now, what kind of computer you would build? I'm facing a similar dilemma, and am fearful of getting ripped off, so I've decided to build my own.
I'm looking for a DVD burner, great video card, overclockable processor, and a decent display.
$US2000 is a pretty respectable amount of money for a home PC assembler, these days. Which is to say, you may find yourself not spending all of it. This is not at all a bad thing, since the depreciation rate of computer hardware is such that it's wiser to invest in, say, beer.
Going through the spec list in order:
Single layer 8X DVD burners are now commodity items, and excellent value for money, as are discs for them. Off-brand 4X discs are dirt cheap, and may be burnable at 8X, not that it probably matters to you much if you have to slow down a bit.
Double layer burners and, most noticeably, their media, aren't yet good value for money. So buy a cheap 8X single layer burner for now, and you can add a double layer burner in another drive bay (assuming you're not building a Small Form Factor box) in a few months, or later if you don't find yourself needing it.
What video card you should get depends partly on what you mean by "great"; I choose to believe you mean "up-to-date, but also decent value". There are a few excellent budget-to-midrange options these days, but the video card to pick depends on the motherboard you choose.
If you're building a whole new box then a PCI Express motherboard, and video card to match, gives you a reasonable upgrade path - but at the moment PCIe also means an LGA 775 Intel CPU, not anything from AMD. The new PCIe gear no longer carries a huge price premium over Socket 478 systems, even if you choose a board that uses DDR II RAM, but the new gear also has no performance advantage over Socket 478, and isn't particularly future-proof. P4s won't be running at a mere 800MHz FSB forever.
For PCI Express, a Radeon X600 or X700, or a GeForce 6600, would qualify as "great" in my opinion; yours may differ. For AGP, the non-GT, non-Ultra GeForce 6800 is still a good proposition.
In the CPU department, I'd recommend you get an LGA 775 system with a mere Celeron CPU and drop in some extra P4 horsepower later, when significantly-faster-than-3GHz P4s no longer cost stupid amounts of money - except for the minor problem that LGA 775 Celerons haven't made it to market yet. The current Prescott-based Celeron Ds (Socket 478 at the moment, LGA 775 Real Soon Now) are quite acceptable at stock speed, can be wound up to roughly 3GHz P4 performance levels - they don't have anything like the nasty cache-and-FSB-based performance loss that the older Northwood-based Celerons suffered. And they're cheap. So it's not completely nuts to buy one for a new Socket 478 system. But if you're thinking about Socket 478, you should also be thinking about Socket 754, AMD's mainstream replacement for Socket A.
The Sempron 3100+ is the standout value chip for Socket 754 at the moment. It's cheaper than the slowest of the S754 Athlon 64s, but it's a merry little overclocker; the consensus seems to be that if you can't get at least a 30% boost out of it, you can consider yourself swindled. Wound up like that, it whistles along at least 90% as fast as a 3.4GHz P4, generally speaking. Even if that 3.4GHz P4 is an overclocked 3.2, it'll still cost more than 1.5 times as much as the Sempron.
If you take the Sempron route and later find yourself wanting a 64 bit chip or (more plausibly) just more speed, you can drop in an Athlon 64 - though not one of the higher spec dual-channel-RAM Socket 939 versions, which're AMD's flagship desktop products at the moment. The single memory channel really doesn't hurt Socket 754 much, though, much to the irritation of AMD's market-segmentation people.
Socket A's still alive, too, but such a system isn't terribly exciting compared with S754 these days. You could save a bit of money by building a machine based on an Athlon XP 2600+ or one of the cheap Socket A Semprons (or save even more by buying one second hand - the Athlon XP God Box of a couple of years ago is today's $US300 eBay special), but you'll only get the stock speed of the faster Celeron Ds out of that, and probably not have a whole lot of overclocking headroom.
On to the display.
Money spent on a nice big monitor is better than money spent on 13% more CPU speed. So let's do some back-of-envelope scribbling. There are a lot of variables, but you should be able to stack up a Sempron 3100+ system with 1Gb RAM, 250Gb HD, single layer DVD+/-R, nice case and plain GeForce 6800 for $US1300, tops. That leaves you $US700 of your budget, which'll buy you a pretty tasty brand name 19 inch LCD screen and leave you at least $US100 change. Or come pretty close to getting you a 20 incher. Or you could kick it old school like me with a 21 inch CRT. $US500 buys a pretty nice one of those these days, probably including delivery for The Carton That Will Not Fit In Many Cars.
(My current Samsung SyncMaster 1100p Plus isn't that big, though; like many modern 21 inch CRTs, its casing is not as deep as that of many old 17 inch screens. It weighs a hummingbird-like 26 kilograms.)
I need to clear/destroy/ruin an old hard drive and storage devices of mine. I read what you said about how the harmful effect of magnets are often hype, but I need to know how to intentionally ruin my hard drive. The size/strength of the magnet and where I could purchase one would be infinitely helpful.
People don't usually bother doing this magnetically - at least, not just magnetically. You can do it with a big-ass coil (the most powerful rare earth permanent magnets might be able to manage it if you took the lid off the drive and rubbed them on the platters, but of course rubbing a rock on the platters would be similarly effective), but Serious Government Agencies generally go for brute physical destruction as well as a giant degausser. Which they and you might as well, because if you manage to degauss a modern hard drive's platters, it will have lost its servo tracks and will now be a paperweight anyway.
Personally, if I had a hard drive full of stuff that I absolutely positively didn't want anyone to see, I'd use one or another "shredder" program to overwrite the data multiple times, which if done correctly is about as good as tossing the drive into a vat of molten iron at stopping Whoever from getting the data back (I've talked about this before). If that didn't sate my paranoia, though, I'd then dismantle the drive (easy to do if you've got a set of small Torx drivers; most drives use those "star" screws), harvest the voice coil magnets out of it (because they're nifty), and do something nasty to the platters - whack 'em with a hatchet, or whatever.
Any chance of you reviewing it or finding out about SATA versions and current prices?
Nope, there's no chance whatsoever.
I can, however, confidently say that there will never be a SATA version of this product, and your price inquiry holds the key as to why.
BitMicro, like every other maker of serious SSDs (most of which, these days, don't use flash RAM; the Mean Time Between Failures figures for the BitMicro products are far better than those for cheap flash RAM devices like CompactFlash cards), is not crazy about quoting prices. This is because their prices are only ever cheaper than refitting a whole server room with newer big iron that can handle more RAM. These things are far, far, FAR out of the range of any consumer who does not also own his own airline. I don't know what the new (relatively) gigantic 155Gb model costs, and it may be significantly less than the $US1000-per-gigabyte that used to be the benchmark for these kinds of things, but I still wouldn't be surprised if American shoppers could get an S-Class Mercedes-Benz for less.
Bear in mind that SSDs that use dynamic RAM are generally quite a bit cheaper than these durable Flash versions, but you're still looking at $US3000 for a 4Gb one of those.
Hence, I doubt BitMicro are terribly keen on the idea of giving me a drive for free.
I have a question about air conditioning in cars and its effect on fuel mileage. I've been told that it reduces gas mileage, and that "everyone knows this". Now, I'm a car guy, and I know something about 'em. The only way I can see you getting reduced fuel mileage is due to the A/C sucking enough power to run that it requires more throttle to maintain a given speed. Sure, on a fairly wimpy car (for instance, our '92 Ford Escort, with a whopping 88HP) it may take enough that it matters, but most modern cars (well, American anyway... our insatiable appetite for large, powerful vehicles never ends) have plenty of power. Our '97 Dodge Intrepid R/T has 214HP, and while you can definitely feel a power loss when using the A/C, I've never felt it was lacking on the highway, nor have I seen the cruise control apply more throttle when A/C was turned on.
Googling brought up mainly rather extremist fuel conversation sites that recommended going so far as to shut off the engine while idling in a drive-through line. Yeah, that'll do wonders for your engine...
A typical automotive air conditioner will rob the engine of between five and 15 horsepower while it's running. This is a pretty impressive mechanical power draw compared with home air conditioning systems, but car air conditioners are smaller and less efficient than home ones, and homes are better insulated than cars. As you say, even 15 horsepower isn't much to lose if you've got a powerful car (or whatever), but the engine will have to turn the compressor when it'd otherwise be idling, when it'd be on the overrun while you slow down for the lights, and at other times when its fuel consumption would otherwise be close to zero. So even a relatively small air conditioner load can add up to a readily measurable fuel consumption difference.
(Note, by the way, that some cars have an automatic air-con cutout relay that turns off the compressor whenever you use full throttle.)
Whether the economy loss from air-con is enough to care about is another matter. Not many cars are likely to lose much more than one mile per gallon when using air-con. Or, in countries like Australia that use the marvellously intuitive litres-per-100km economy metric, maybe a whole litre per hundred klicks, for a super-economical car; big family cars might not even suck down another half-litre per hundred kilometres. If you're driving a one litre Daihatsu then this can mean 10% more fuel consumption; generally speaking, the more fuel your car normally consumes, the smaller will be the proportional impact of air-con.
Also note that there are lots of other things that can cause a similar, or worse, fuel economy hit. Accelerating hard up hills, for instance, is a great way to drain the tank. Driving with your windows open rather than the air conditioner running can, if you're going fairly quickly, cost you more in drag than you'd pay to run the air-con. And, of course, people who drive around on underinflated tyres with a bunch of heavy crap sitting in the back of the car are likely to pay a lot more for all that than they would for occasional air-con use.
Of course, in getting "gaming" computers, you are normally paying for the idea, as the component parts are almost always far cheaper to do yourself, or even get from another customizeable system seller. But still.
Not fabulous value, no, particularly seeing as the only quantifiable difference between a Colonial-Marines-spec cable like this and a regular $2 IEC lead, in normal service, is that the more expensive one may be heavy enough that it wants to unplug itself all the time.
Now, the three 12AWG conductors in that cable are impressively chunky, but they're hardly necessary.
I just grabbed the nearest IEC lead and read the stuff printed down the side, which told me the conductors inside had a cross-sectional area of one square millimetre each; that puts them around 17AWG - actually, they're probably rounding down from the 1.3 square millimetre area of 16AWG, because that's a standard size for light duty power cables, and 17AWG isn't. Anyway, 16AWG wire is what you'll find in every standard ten amp light duty extension cord.
A ten amp 110VAC cable can run an 1100 volt-amp load. That may be only about 800 watts worth of PC equipment, thanks to the joy that is power factor, but it's still plenty more than you'll need for a PC and its monitor put together. 220V countries get more than twice the power transmission out of a cable with a given current rating, so 16AWG is even more oversized for the PC-powering job there. Here in Australia, our nominal mains voltage is 230V, so in theory we can make do with even thinner wires.
12AWG super-heavy-duty cords have a standard 20 amp rating - so the 110V Alienware one is as over-the-top for powering PCs in the USA as a $2 lead is here in Australia! Hurrah!
The Alienware lead also, of course, does not deliver the promised "ultimate conductivity", unless it's a room temperature superconductor. It does, of course, have lower resistance than a 16AWG lead; the resistance of eight feet of 16AWG (the same length as the Alienware cable) is 32 milliohms, while eight feet of 12AWG is a mere 13 milliohms. Take into account the high-spec plugs on the ends of the Alienware lead (...which have to mate with the same loose and corroded copper sockets as any other lead...) and the cable resistance may be dwarfed; who knows, the Alienware lead may deliver a whole tenth of an ohm less resistance than a cheap lead!
Assume an implausibly high 500 watts is being drawn by a super-stacked PC and its big-ass CRT monitor through this cable, and ye olde formula V=IR tells us that the effective resistance, ignoring power factor shenanigans for now, is more than 24 ohms (and the current's about 4.5 amps, less than half 16AWG's rating...). A 0.1 ohm supply cable difference will therefore have a total effect on the voltage the PSU sees of less than half of one per cent. And the less current the cable's passing, the worse it gets; at 300 watts (still more than most PCs, even tweaky ones, draw by themselves), you're talking 0.25%.
PSUs are made to seamlessly handle the normal swings, surges and spikes of metropolitan power grids, amounting to at least several per cent either way; using a cable that gives them a whole 1% less voltage really isn't going to faze them.
Heck, Australia switched from 240VAC to 230VAC a while ago - permanently dropping the nominal mains voltage to 96% of what it was before. Most people didn't even notice.
But, y'know, the Alienware cable does have that "mesh shield overcoat" to "prevent interference". Those of us who habitually drape our PC power cables over the nose radar of a MiG-31 will be sold just on that feature.
There is a company called Better World Technologies, which seems to work closely with United Community Services of America. They claim that they can generate 5 times as much energy as is put into their generator. I personally find it quite difficult to believe, however, I wasn't able to put a hole in their science.
My friend brought home a video from school that shows how their device works. I watched it with him, of course being very skeptical. Afterwards, I looked it up online and found various websites. Considering the video was put out in 1996 and they claimed to be close to production then, and still claim the same thing today, it's not good for them. anyway, you've examined this type of thing before. could you take a look at it for me?
For a person who's spent so many years being totally dedicated to getting free energy knowledge into the hands of the public, and who's taken so many deposits and made so many tours, it's amazing how Dennis has never managed to give one of his miraculous machines to any independent tester to see if the damn thing actually works. It's just that dog-and-pony stage show, and then the amazing never-to-be-repeated opportunity to "invest" your life savings, every time.
I've mentioned Dennis in passing before.