Dan's Data letters #91
Publication date: 15 February 2004.Last modified 24-Dec-2012.
Bespoke computing?
There has been much speculation about a company supposedly making really fast computers. Do you think you can shed some light on this subject. The name of the company is Liebermann Inc; their Web site is here.
Some of the specs they list seem to be out of this world, especially about the a technology called "PuRam", which no one has been able to test. Let me know what you think.
Igor
Answer:
I think we have a serious contender for
this year's Blatant Apple Look And Feel Rip-Off Award!
Anybody who spends an extra $US4999 to buy a PC from Liebermann can test "PuRam"; that's the cost of a PuRam drive for one of their PCs. The PuRam is just a solid state drive of unspecified capacity (don't expect much), and $US5000-odd is only a slightly inflated price for one, by market standards. Those things are hilariously expensive, though there's no terribly good reason for it these days (I've written about SSDs before, here).
The rest of the Liebermann PC features are plausible enough, including the clock speed, because these PCs are built in Asetek Vapochill cases. The 3.8GHz system they've got as their current top-of-line desktop box is a perfectly achievable overclock with a superpowered CPU cooler like the Vapochill or Chip-Con Prometeia (now known as the, um, nVENTIV nVENTIV).
Anyway, all vapour-phase cooled PCs mean you're absolutely paying through the nose to get the kind of speed that you'll be able to get with air cooling in about a year, tops. But the Liebermann ones look like nice enough computers. You could put the same thing together without all of the fancy trademarked names for rather less money, of course - a LOT less money, if you're willing to give up a few hundred megahertz, the solid state boot device, and the stone logo.
[Or, then again, perhaps it's all a scam.]
Sparks and boxes
Can you recommend any basic texts on electronics, wiring, and similar things? I'm interested in starting to fiddle around with electronics, but I really don't know where to start. Do you have any suggestions?
Oh, and as a side question, I'm upgrading my computer case. Are there any cases you've reviewed off-hand that you'd recommend over a Lian Li?
Nick
Answer:
There's plenty of good electronics starter info on the Web.
This site, for instance.
You may also find this site and this one helpful when it comes to connecting this to that.
Once you've got an idea of the rudiments, head off to your local place-that-sells-electronic-kits and browse their basic, educational range. There are plenty of things with circuit boards the size of a playing card that'll teach you about elementary circuits, and they don't have to be useless "Thing That Goes Beep" and "Electronic Dice" sorts of projects; a small Jacob's Ladder is a pretty simple circuit, and very pleasingly dangerous!
Regarding case quality - no, I don't have any recommendations above Lian Li, although boring beige steel cases from various manufacturers are better value, if you don't care about style or an extra kilogram or three.
In the same market segment as the Lian Lis, Cooler Master's aluminium fancy-cases are up to Lian Li standard, and there are some fancier cases out there, sometimes with auto-gloss finish, that make Lian Lis look cheap. But if you can get Lian Lis where you live, you'll be happy with one the right size for the stuff you want to install.
Hype?
All other things being equal, how should the speed of a 3GHz hyperthreaded CPU compare to a dual-1.5GHz SMP setup, with more than one CPU-heavy program running?
Bennett
Answer:
It should compare very well, but not nearly as well as Intel would like
you to believe.
Hyper-Threading (in popular usage, it seems to have lost its hyphen...) is technically interesting, by and large, a gimmick. The system appears to have two processors, but they're two processors that're so seriously bottlenecked by sharing the same cache memory that for most practical intents and purposes they're only one, running at the specified speed.
Hyperthreading seldom hurts performance, but a couple of real 1.5GHz CPUs should, if you're running lots of tasks, beat a single 3GHz HT CPU by an easily measurable, and possibly even noticeable, margin.
Camera choice #376
I am going overseas soon and want to buy a digital camera, but the more I research and read user/pro reviews, the more confused I become about the conflicting info.
Some say the Canon Powershot A80 gives a "soft" image, yet others say the focus is sharp. The Olympus C-5000Z gets a good writeup on some sites, as does the oldish Powershot S50. And so on for the Coolpix 4300.
What would you buy, given:
* family/tourist type shots with a desire for some control over shutter speed and aperture at times
* sharp, sharp shots
* medium sized body or lessAlso, do you think the disproportionate price jump between 4MP and 5MP (when compared to increments from 2 to 3, 3 to 4, etc) is worth it? Can you recommend a good review site?
Russell
Answer:
Different digicam lens/sensor combinations give different sharpness per
pixel, all things being equal; you can't expect an ultra-compact four megapixel
camera to produce results as sharp as a prosumer model with the same resolution
but a much bigger lens. You also can't expect a consumer camera with a ton
of optical zoom to produce results as sharp and undistorted as those from
a lens that costs more but doesn't have as much zoom.
That said, though, I haven't heard any complaints about the Canon A-series cameras producing particularly soft results compared with competitors in their class.
Sharpness isn't the be-all and end-all, anyway. A lot of photographic enthusiasts get very hung up about sharpness (and "accutance"; find a primer on the subject here), but a great photo isn't necessarily one where every detail jumps out and bites you on the retina. Don't freak out about sharpness in particular.
A quality lens does make a big difference, though. Many digicam buyers fixate on resolution, assuming that a five megapixel camera must create more detailed pictures than a two megapixel one. Well, all other things being equal, then probably yes, but a subcompact 5MP camera with an inescapably fuzzy lens can produce much less impressive A4-sized prints than a 1600 by 1200 camera that was the New Prosumer Hotness when it came out.
Regarding your "medium sized" criterion, that depends on what you mean by "medium", but if you just mean "not something you can forget you put in your pocket", there are still plenty of options. I'm not so sure about the C-5000 (only because I don't know much about it, not because I think it sucks), but the other cameras you list would probably all suit you.
Regarding price-per-megapixel - I'd pay for extra resolution in exact proportion to how much there is - I'd pay 1.25 times as much for a (genuine) 5MP camera as for a (genuine) 4MP version, assuming the lens could deliver all of that resolution. Which it can't, necessarily; Casio's tiny and excellent Exilim EX-Z3 is a better buy than the higher resolution EX-Z4, for instance, because the Z3's sensor is already pretty much at the limits of the lens's capabilities).
For digicam reviews, the inventively titled Digital Photography Review is a one-stop shop for any camera it covers, but it doesn't have reviews of everything. Steve's Digicams is also generally very good, as is The Imaging Resource.
Yo-yo power
Here's an interesting site that came up during discussions about efficient vehicles on the Toyota Prius Yahoo Group.
It seems to describe a single-cylinder motor/generator which uses a single piston with combustion chambers at either end of the cylinder (in essence, somewhat like a flying shuttle loom). The idea being that by putting a combustion chamber on either end of the piston, you fire the piston (containing another coil) back and forth through a wound coil to create electricity for cars, boats, nuclear submarines, etc.
Now (in the event they're actually serious about this) it seems to me that most of the power being used by the combustion process would simply be lost in changing the direction of motion of the piston inside the cylinder, before any energy is actually created. Not to mention that a conventional multiple-cylinder engine powering a conventional rotating generator would seem to be more efficient with less wasted energy.
Your thoughts?
Dave
Answer:
Is Red Barn Engineering serious about
the "Mogen"? Well, it doesn't look as if they've actually built one
yet, so who knows.
You may be right about energy loss, but I think with appropriate valve timing the piston can be "braked" by doing the work of compressing the fuel-air mixture at that end of the cylinder, and then the spark plug can fire when the piston's on the way back.
This creates some interesting problems of its own, of course.
In a conventional piston engine, the fuel/air mixture is (or at least should be) ignited late enough that the piston's pushed down by the expanding combustion gases after the crankshaft has passed Top Dead Centre. Otherwise, as I'm sure you know, you're asking for engine damage by pushing the crank against its rotation.
The double-ended piston idea has no crank and so doesn't have this problem; there ought to be very wide latitude for ignition timing, with better timing (presumably servo-synced to the piston oscillation speed, a la brushless electric motors that energise their drive coils when they know the rotor's in the right place to be moved by them) just yielding better efficiency.
This is all very well in theory, but getting the compression and spark timing right might well be rather excitingly difficult, and firing too hard from one end could send the piston back the other way hard enough to cause detonation, just like what happens when you run a normal high-compression engine on low-octane fuel.
The amount of magnetic braking the piston/armature assembly would experience as it shot back to the other end of the tube would also vary depending on how much load the generator was under. There's another variable.
Perhaps all of this could be handled easily enough by a simple electronic control system or even a mechanical governor of some sort, but I wouldn't be astonished if it turned out to be an unstable system that could oscillate itself into destruction.
Less peculiar spring-loaded and crank-equipped versions of the idea are shown on this page, mind you; just sticking a crank and flywheel on the end of the piston shaft would make any and all timing and travel-limiting and compression nightmares go away. But, as you say, if you're doing that then you probably might as well be using a rotary generator, unless the linear generator idea is a great deal more efficient.
Chips continued
A few readers sent me their thoughts about my comments regarding "hot chips" for cars in the last letters column. I was also struck a glancing blow by Slashdot. Here's the correspondence I received.
Terminal T-pieces
After reading letters #90, I found an article of yours that I hasn't read (the horror!). So, I promptly started to read it, and was quite surprised when I found this statement:
"On turbocharged engines, hot chips can also increase the manufacturer's boost limit settings, so the turbo boost can be cranked up. This can give hefty performance increases, especially on turbo diesel vehicles that are more conservatively tuned by default. Of course, overpumping your turbo can also wreck it, and more cylinder pressure is another thing that can encourage pinging. And you can gain a large portion of this performance gain more cheaply, by leaving your ECU as it is and simply installing a T-piece in the pressure line running to the turbo wastegate, so it opens later."
The part that surprised my was your (more or less) recommendation of using a T-piece to fool the pressure sensor instead of buying an expensive EEPROM upgrade. This is not a good idea! Modern ECUs are able to compensate for this "hack", heck, some cars will even indicate engine problems. And even if they don't, you run a higher risk of damaging your engine/turbo/whatnot in some way, since you don't have any idea if you are getting an 0.5bar or a 1.5bar increase.
How come you know this, I hear you ask... Well, I tuned my last car by the means of an "optimized" ECU. But this upgrade was not made without some research (a lot actually).
But you're right about the specs ... They are quite "optimistic"! Although my Volvo S40T4 did get quite a lot more fun to drive -nothing like spanking some BMW booty when they aren't expecting it!
If you're wondering, the factory specs for a Volvo S40T4 (high pressure turbo engine) is about 200hp and about 300Nm of torque, and according to the tuners, the upgrade should provide about 230hp and about 360Nm.
(Plain chip swap lousy value for money? Cars are lousy value for money, i.e., black money absorbing holes.)
Henric
Answer:
Fair enough about the T-pieces. When I wrote that piece, none of the rev-head
friends I consulted in the course of my research could afford cars smart
enough to compensate for such a hack! I've amended the article appropriately.
Rule-proving exceptions
I just your 90th letters column and noticed the discussion on engine upgrade chips for "improved performance". Needless to say, I agree with the majority of what you had to say regarding the subject. However, there is at least one exception to most of the rules about these chips.
In the North American and European markets, many turbocharged Volkswagen and Audi cars use an electronically controlled wastegate to regulate boost levels. I'm not sure if those cars are configured the same way in Australia, so it may not apply in your area. On those models, a simple chip upgrade can give pretty significant bump in torque and power by increasing the allowable pressure developed by the turbo.
The stock boost level is (I think!) .5 bar, and chips can increase it to .8 bar or even up to 1.1 bar. The overall effect is similar to installing a boost controller, in addition to the other minor changes like fuel map tweaking. Of course, there are still limitations to the extra power you can get with this; air flow can only be increased so much before you go outside the optimum compressor efficiency range.
Even so, increases from 170 hp (~127 kW) to about 205 hp (~150 kW) can be achieved in my model of car (VW Passat with 1.8 liter turbo). So a 15% increase is not an unreasonable estimate in this specific case. I've seen plenty of discussion on engine chips in the forums at Club B5 and VWvortex. People have also posted actual dyno test results verifying the positive results.
For most cars, I know the increases are much, much smaller. I think that, in typical marketing fashion, engine chip manufacturers are using this one specific case as the upper range of the improvements they quote. Of course, the fact that it doesn't apply to the vast majority of cars doesn't deter them. So it's highly misleading, but not technically a lie. Anyway, I thought you might not be aware of this one data point that may explain (but not excuse) some of the manufacturer claims.
Now, just to cover the bases: I'm not a technical expert on any of this stuff and I'm not involved in any way with any aspect of the auto performance industry. I'm just a guy who likes his VW!
Brian
No replacement for displacement
Regarding the discussion on Powerchips, I went through the initial enthusiasm of "Wow... my car will fly off the road with these things!" a few years ago, being a kid with his own car (and able to Frankestein the thing without my father Frankensteining me). At the time, we were all crazy about tuning our cars in some way, so we spent half the time at this one garage owned by a friend's father that would give us nice discounts.
I brought this up with the mechanic there and his answer was basically "ummm... no". His real answer was that these chips only have a real effect on large engines, not the dinky 2-liter in my 306. He said that smaller engines (up to around 3.5L) are already optimized heavily and usually have the best settings possible electronically, the only upgrade able to make a difference is something mechanical. Now, larger engines are usually tuned down electronically, in order to meet emissions controls and compensate for the crap gasoline people put in them, so for those engines, the claims made by the chip merchants are "more true".
Nevertheless, he agreed with you that, for example, a larger air filter with a shorter loop would be much better than an ECU upgrade.
By the way, I ended up getting the chip anyways (damn stupid kid).
Hugo
