Atomic I/O letters column #77
Originally published in Atomic: Maximum Power Computing Last modified 16-Jan-2015.Grease me up, woman!
I'd like to submerge a PC in liquid. I've read about it online and also know about Atomic's own mineral oil PC - my question is where can I get the oil?
Most projects of this nature use vegetable oil, but it doesn't look particularly good and tends to break down. I know there are other products available, but I don't want to resort to selling organs to cover the cost.
Samuel
Answer:
As you say, vegetable oil is a bad choice. It goes rancid when exposed to the air, and
will do so faster when kept warm by a computer.
Mineral oil is good, because it's transparent, not very expensive (unlike silicone oil, or really exotic liquids like liquid fluorocarbons), and won't break down. If you don't care so much about appearances, light motor or lubricating oil should do the trick too; that's pretty easy to find, of course.
(Glycerol is not a great idea, by the way, partly because it's more expensive than mineral oil but mainly because its hygroscopic - it'll soak up water from the air, and slowly become more and more conductive.)
You'll need 15 to 20 litres of oil for the average submerged PC. To get that much clear light mineral oil, you can either buy 150 to 200 of those little bottles of Singer sewing machine oil from the supermarket, or you can phone... a large animal vet.
Clear light mineral oil, you see, is used as a laxative.
A few spoonfuls of it will probably have a... noticeable... effect on a human, but you need quite a bit more if you want to get a horse moving.
So vets who treat horses will probably be able to hook you up with a supplier of gallon-and-bigger jugs of just the oil you want.
(After this page went up, a reader pointed out that industrial suppliers, particularly those that deal with industrial sewing machines, may well have access to large containers of mineral oil. Oh, and baby oil is also likely to be pretty much straight mineral oil, plus a little fragrance.)
Ice... from fire!
Recently I've been considering joining in on the fun that blacks out many regions of Oz over summer by purchasing an air conditioner.
Although I'm quite happy to pay the bills just to keep my room and PC chilled, I've been hassled by various family members about the costs. The thought of rigging up a few solar panels to the air conditioner crossed my mind, but common sense tells me I'd only have enough juice to cool my outback dunny for a couple of seconds.
Could I actually run a small air conditioner with solar panels, or should I just strap a few heat sinks to my nether regions to complement my body piercings and stick it out through the heat?
K.
Answer:
You'd have to spend a lot of money to get enough solar panels to run an air conditioner.
A small window air conditioner will draw something in the order of 1400 watts when its compressor's running - rather less when the thermostat's clicked off and it's only running its fan.
Solar panels are rated according to their maximum noontime direct-sun output, so you can't run a 1400-watt load from only 14 "hundred watt" panels. Realistically, even if you're only interested in running the air conditioning during the day and so don't need more than a small battery to smooth out panel output on the way to your inverter, you're going to need more like 3000 "watts" of panels to run a little 1400W air conditioner more or less constantly.
Tracking hardware that turns panels to face the sun can considerably reduce the wattage you have to buy, but it increases the price and hassle of the installation, and doesn't do anything to prevent trees and clouds from blocking the light.
The retail price for new solar panels at the moment is an easy $AU7000 per kilowatt (and you need about eight square metres of space per kilowatt, too). So the whole system could easily cost as much as a new car.
Interestingly, though, there are solar air conditioner products out there - generally hybrid systems that have a mains power connection to make up whatever portion of the input power the solar system can't supply. If they're normal air conditioners then they won't be able to keep on operating without the abovementioned large and expensive solar array (upcoming low-cost solar technologies may do something about the price, but they're lower in efficiency and so need even bigger arrays...), and they'll need a monster battery bank to work without mains power at night.
Most of the hybrid products don't seem to actually work that way, though; they just combine an evaporative (or "swamp") cooler - that just blows air over a water evaporator - with an air conditioner. Because the evaporative cooler only needs power for a fan and perhaps a small pump, a small solar array can run it with no trouble.
If you live somewhere inland with low humidity, you can cool your house quite well by using an evaporative cooler. Evaporative coolers can't drop the temperature of the air by a great deal, and they're useless in medium-to-high humidity places (Mount Isa, yes; Brisbane, definitely no), but they're very cheap to run.
(Which, of course, means that running one from solar power instead of the mains won't actually save you much money.)
And now, finally, the option that would actually work and not cost a fortune, but which you can't buy off the shelf: An absorption refrigerator.
You don't need electricity to run a heat pump. As the old kerosene-fueled absorption refrigerators (and the very entertaining Crosley IcyBall) proved, all you need is a heat source. That heat source could very easily be a simple solar-reflector-and-black-tubes arrangement, heating the absorber section of a good old fashioned ammonia-based refrigeration system.
Ammonia systems have the disadvantage that they use not just ammonia but pressurised ammonia, which is nasty if it leaks. But they're reasonably efficient, and run from heat directly. There's no need to bring expensive, inefficient solar cells in on the action; a load of pressure tubing painted black will do.
Most absorption refrigerators, like the IcyBall, are manually switched from cooling mode to regeneration mode. You heat one bit of them, then when you let the device cool, another bit gets very cold. In the old kerosene fridges, this was achieved with a small kerosene reservoir; you filled it and lit it at regular intervals, and when the fuel was all burned, the fridge chilled its interior.
Here's a page about this idea applied to air conditioning.
Big slots, little slots
Will a PCIe x1 TV tuner card (Kworld PCIe Dual Hybrid TV Tuner), or any other x1 card, work in an x16 PCIe slot on my Asus Striker Extreme motherboard? I have two video cards and a Creative X-Fi installed already, and my only free expansion slot is the x16 slot.
I would prefer to use an internal expansion slot rather than buy a USB TV tuner, because the bandwidth of USB is much less than the PCI bus.
Geoffrey
Answer:
Yes, it'll work. The little connector on PCIe x1 cards fits into the matching part of
the longer x16 connector, and away you go.
Note that some slots that look like x16 actually aren't. It's normal for older dual-x16-slot motherboards to only have eight PCIe "lanes" connected to the second of those slots; that works fine for SLI/CrossFire, but if you've only got one video card and you plug it into the "x8" slot by mistake, you'll lose a bit of performance. An x1 card should work just fine in the x8 slot, though.
On your board, you've got two blue full x16 slots, and one white eight-lane x16 slot between them (along with an x1 and a plain PCI slot).
More perniciously, some motherboards have an "x16" slot with only one lane connected. They're usually boards with onboard video; the manufacturers don't expect many customers to use the "x16" slot for a video card, so they cut corners and save a few bucks.
Some, but not all, x16 cards will work in one of these cut-down x16 slots. Once again, though, any x1 card should be fine.
Ctrl-Vefrag
I do a lot of encoding whenever I record some videos, and I find that my disk fragmentation level becomes very high within three months.
I am planning to build a HTPC containing a fair amount of storage (500 Gb+). I'll be heavily investing into more space for this upcoming project of mine.
Rather than doing more frequent defragmentation, I was wondering if it's faster and more efficient to create smaller partitions, say 100GB each, and schedule a task which will do the following every 3 months for each partition:
1. Copy the contents of a partition to another empty partition
2. Format the current partition
3. Then copy the content back to the original partition once the format has been completed.If there was a partition that is very fragmented, would copying all the files to another partition automatically resolve fragmentation issues? Would copying the files reallocate all of the data blocks contiguously, or will it duplicate each block like an image?
King
Answer:
Copied files are indeed laid down in contiguous slabs, if the empty space on the destination
drive permits it. They're not laid down in any particular order - modern defrag
utilities put oft-used programs on the faster
outer tracks of a
drive, if possible - but copying data to an empty drive or partition will indeed defragment
it in the process.
If you're copying between physical drives, this should also be quite a lot faster than defragging one drive. Defragmentation flogs the drive thoroughly, reading from here and writing to there zillions of times; it tries to speed the process up by using some of the drive's empty space for temporary storage (which is why defrag utilities complain, and run slowly if at all, if you run them on an almost-full drive), but there's only so much you can do when you're shuffling data around on the one physical disk.
If you're moving data between partitions on the one disk, though, the speed advantage will be much less impressive. If you just copy the data from Partition A to a freshly-formatted Partition B on the same drive then the process may be a bit faster than defragging. If you then copy back from B to A again, as you're proposing, both copy operations together are practically certain to be rather slower than a defrag of a single larger partition would have been.
Note, also, that many PC enthusiasts defragment their hard drives much more often than they need to. If you're doing stuff like video editing that needs lots of huge sustained transfers then fragmentation will have considerably more impact, and defragging every three months isn't excessive at all. But remember that back in the day, Microsoft advertised the NTFS file system that's standard for modern Windows machines as being so speedy that it didn't need defragging, ever.
That wasn't, as it turned out, true. But NTFS certainly does deal much better with file fragmentation than bad old FAT32, and many people still used to defrag their FAT32 Windows 98 computers much more often than was necessary.
Like magic, I ruin your credit card!
At a party, someone told me that he heard of a man who had strong magnets implanted in his fingers. The purpose being, to detect magnetic fields. Apparently the magnets were coated in silicone, and that this allowed for nerves to grow around the magnets, therefore becoming part of his nervous system. When waving his hand near a magnetic source, he could 'feel' the magnetic field.
My question is, can he really feel it? or is it just the feeling of a magnet trying to pull itself out of his skin? And where can I find a doctor crazy enough to give me some magnets of my own?
Steve
Answer:
Quite a few people have done this over the last several years. It's not a difficult
procedure, but apparently they haven't quite gotten around the problem of the magnets
breaking down and/or being rejected by the body.
As you say, they have so far been using tiny rare earth magnets with a silicone coating, but that's because the silicone is inert in the body, not because it encourages anything to grow around it.
People with these implants certainly can feel the magnet move, though; you don't need some special nervous connection to be able to feel very clearly everything that a rice-grain-sized thingy embedded in the end of your finger is doing. The usefulness of this new sense is questionable, but it works just fine.
Unfortunately, the silicone coating breaks down over time, and then everything goes to hell and the nasty remains of the magnet have to be dug out again. This is obviously a solvable problem; the magnets just have to be coated with something similarly unreactive but more durable.
If you want to join the few, the proud, the people who can't have Magnetic Resonance Imaging done on them for a very unusual reason, then there are probably plenty of piercing shops that can help you out. Freaky people have been getting subcutaneous implants of various sorts for ages, and the finger-magnet one is a very minor operation compared with your average Klingon ridge job.
Some relevant links (Warning! Grody pictures!):
So what's it like
having magnetic vision?
A Wired article on the
subject.
The gruesome aftermath.
Steve Haworth's
site.
And, as a bonus: the famous
Eyeglass Piercing!
