Dan's Data letters #184Publication date: 14 April 2007.
Last modified 03-Dec-2011.
After several years of wishing that Total Annihilation would improve, I read with interest your writings about Supreme Commander. Fantastic screenshots, and wild praise abound in your prose about everything from visual stimuli to gameplay.
You sold me. I bought the game.
However, I find performance lacking on my computer. Even with the quality settings all set to minimum, the performance is often very poor. And so, while it had seemed to be a perfectly adequate machine for every other game I'd ever thrown at it, Supreme Commander has shown that things are no longer sufficient.
So, I dare ask: What are you using to play this game? I sense certain upgrades in my future, but require direction as to what they might consist of. And I'd certainly prefer to hear that you're running the game with DDR RAM and an AGP video card, as that would obviously be the most economical upgrade path for me.
Don't worry - everybody playing Supreme Commander finds performance to be "lacking" on every computer.
SupCom, like TA when it was new, is just a really demanding game - though TA was 100% CPU load, while SupCom grinds both 3D cards and CPUs. Future patches may optimise it significantly, but this is not like the early release of TA: Kingdoms, which was painfully slow on account of inadequate development time. SupCom's just a giant.
Fortunately, though, it's not actually necessary on many pretty-current computers, to turn all of the game settings all the way down. No current computer will play a big game of SupCom smoothly at normal game speed no matter what you do, but you can fiddle with rendering settings separately to get a decent frame rate on smaller games without turning off all of the pretty.
Basically, graphics card power determines the resolution (including second monitor, if you have one) and prettiness settings that'll ever give you Frame Rate X in even the simplest SupCom games, while CPU power determines how much the game slows down when there are tons of units doing things. The CPU's the limiting factor in busy games, so there's no harm in turning the prettiness up then, if you've got a somewhat recent graphics card; you might as well see ground decals and glow effects while you wait for the CPU to figure out where all of those siege bots are going.
Note that the keypad slash key turns on the frame rate counter, for quick checks.
My computer is nothing all that thrilling. It's basically still this machine; Athlon 64 X2 3800+ at 2.2 to 2.5GHz (depending on how this poor abused processor feels...), GeForce 7800 GT slightly overclocked. That's good enough for one 1600 by 1200 and one 1024 by 768 monitor in small to medium games without turning the speed down massively, especially if you turn off decals and bloom.
Note that SupCom, as it stands, only gets significant benefit from a maximum of two CPU cores. More cores are not useless, but it's the first two that work on the simulation thread and the rendering thread, and those are the 800 pound gorillas. It's not inconceivable that either or both of the big threads could be further split in future patches, but I wouldn't bet on it.
So there's justification for a dual core CPU, but not for a quad core. It's also questionable whether there's any point to having a lot of RAM, though you'll be able to start the game faster, and switch from it to Windows and back again more easily, if you've got 2Gb or more. Otherwise, Windows will have to do more paging, and you'll be waiting longer - but normal gameplay should still be fine with 1Gb, if you don't have some other monster app chugging away at the same time.
On your AGP machine, a 256Mb or maybe 512Mb AGP Radeon X1950 Pro card would probably be your best bet. Quite fast, and not stupidly expensive for what it is; here in Australia, Aus PC Market currently have an HIS-branded 256Mb AGP X1950 for only $AU280.50 delivered (Australian shoppers can click here to order!). That's not peanuts, but it's also not at all a bad price for the boost it can give any old AGP machine. If you've got a 1GHz Pentium III or something then such an upgrade is a fairly goofy idea, but if you've got a 3GHz Pentium 4, you may find it'll tide you over for quite a while yet.
Three-point-something gigahertz P4s are pretty much the pinnacle of AGP PC performance, though; you probably can't get a bodaciously fast CPU, by modern standards, for whatever your motherboard is. But CPU speed has been plateauing off anyway, so it's up to you to decide whether it's worth the whole-new-computer upgrade to an AM2 or LGA775 system at the moment.
Barring magical gifts from game patches, it'll be a few years before we can run big SupCom games smoothly on any computer, and by then all of the current standards will be obsolete. So you might as well resign yourself to running with tweaked settings at lower game speeds for the time being.
(-4 speed still gives you quite a lot to do every real-time second in a big game, anyway!)
Why is it so hard to get high dot pitch LCDs?
My IBM T40p laptop has a 14 inch 1400 by 1160 screen.
Trying to find a desktop monitor with resolution that high without going to 20 inches or more is impossible.
Trying to find equivalent laptop screens is nigh-impossible.
I don't want inches, just dots. 1600 by 1200, 10 inches, would be just fine.
Therefore I have to assume that laptop screens are higher quality than desktop. Screen dot pitch also seems to have gone backward in the last few years. Many 13 and 14 inch laptops are now 1024 by 768.
There's nothing in the world stopping monitor makers from putting top-resolution laptop panels in desktop monitors, or from putting top-res panels in more of their laptops. The result wouldn't be a lot more expensive than lower resolution screens of the same size, either.
I think there are two reasons why it doesn't happen.
One: The great unwashed monitor-buying public buy monitors by size and price, not by resolution. A $100-more-expensive 17 inch screen with twice as many pixels just won't sell to most of the market. The same applies to laptops, to some extent.
Two: People sit further away from desktop monitors than from laptop screens, and the operating systems most people use today (or, more accurately, the applications running in those operating systems) don't let you scale stuff up so you can bloody see it at very high resolutions.
The problem, which I come face to squinty-eyed face with every time I use my sister's Inspiron 9200 with the mongo-bodacious maximum res screen option, is that Windows still can't scale everything up properly. Very close to none of this is Microsoft's fault; it's just that too many third party apps (and Web sites, for that matter) have bitmapped elements in them that can't be made any bigger. So you end up with teeny-tiny window gadgets, or text, or whatever. If I were still 17 years old this wouldn't be as much of a problem for me, even at desktop distances, but it'd still be a problem for the average grown-up.
I dare say that eventually we'll all be using 300dpi screens with fully scalable everything. But we're only inching towards that goal at the moment.
Tonight I knuckled under and performed a battery swap on my aged-but-still-jammin' iPod Mini. Two screws, one wire connector, done in five minutes. I'd been liberal with my schedule and allowed myself about two hours for the project (call me butterfingers...), so I decided to disassemble the failed battery to see the magical hamster wheels inside.
The battery was a sealed metal can, and poking a hole in it revealed scary bits of a black substance, plus some coils of copper wire. The thing is, when I punctured the can, a pleasant, dead-on perfect bubble gum scent came out, strong enough that I could smell it from a distance of several feet (the stuff inside didn't taste anything like bubble gum).
So what's with that smell, anyway?
I bet you were smelling lithium. Plain, straight, metallic lithium does apparently have a distinctive sweet smell, and so may the lithium-salt electrolyte in a gadget battery. Said smell may be noticeable near crashed high performance electric model aircraft, and experimental electric cars whose drivers are hightailing it past you in search of a bucket of sand.
I haven't smelled it myself, though, because I haven't yet gotten around to dismantling any non-rechargeable lithium batteries to salvage the metallic lithium in there (the instructions for doing which usually tell you blandly that the lithium will probably catch fire at some point...).
I'm looking for a screwdriver that can extract a screw with a triangular recess. Like an Allen head but triangular. I was hopeful when I heard about a "Gamebit Driver", until I found out that that won't do the trick.
Could be a "TP3".
A Tri-Wing bit just the right size would probably extract it quite easily, in the same way as Torx ("star") bits fit fairly well in hex screw heads ("gamebit" screws are of the "external star" type, like a negative of a Torx screw - it might actually be possible to use some Torx screw heads as external star drivers!). But otherwise, nothing but the official driver will fit.
If the screws aren't in a hole (or if they're only in a shallow hole and you don't much care about the surface around them...) you can cut a slot in them with a Dremel emery wheel, and then use a flathead driver. You can also, if you're courageous, blob some solder or epoxy into the recess and stab a suitable flat driver into it while it's still liquid, then hope it'll hold well enough for you to get the screw out when it's set.
(About ten minutes after this page went up, a reader suggested the time-honoured strategy of heating up a ballpoint pen tube with a cigarette lighter and pressing it onto any weird screw to make an impromptu driver. This technique can even work pretty well on those "one way" screws with a sloped recess that engages the driver when it's turning one way, but slips when turned the other way. I think it's perfectly fair to remove any such screw, whenever you see it, even if you personally have no interest in what's underneath.)
I'm looking for a way to identify some prankster who's sneaking into our backyard late at night and messing with our pool's filter timer (setting it to go off at 4AM and the like).
The idea I've come up with is to use my cheap USB webcam, a couple of IR LEDs and a video capture program set to go on motion or something. There's a pipe through the wall just above the filter (it's next to the laundry); I'm planning to leave my laptop inside against the wall and run the webcam/light cables to it through the pipe.
Illumination is the problem. I bought a couple of IR LEDs to play with - two 5mm 1.2-2V - and experimented with them and a couple of AA batteries, but they don't actually seem to put out a lot of light. The webcam definitely sees them if they're pointed straight at it, but using their light it could barely discern the outline of my hand from about 12 inches away.
It's going to need to light up a face a good metre or so away, and it'd be convenient if the light ran from USB power too.
Should I just go buy a bunch more of the same LEDs? Are brighter ones available? Or should I just go buy an IR Photon light, or a cheap webcam with built-in IR lights or something?
Thanks for any advice (I'm pretty hopeless with basic electronics).
Resetting your pool filter timer. That's... different.
Anyway, you'll need a considerably more powerful IR illuminator than a couple of LEDs. The illuminators ones built into some webcams probably won't cut it either - they generally only have single-digit numbers of LEDs, and the total power available from USB to drive them is 2.5 watts, which has to power the camera itself too.
Commercial 12 volt IR illuminators with 140-odd LEDs in them are down to $US60 or so delivered on eBay these days. They draw about 12 watts, and output plenty of light for use in various surveillance applications with only one illuminator.
Alternatively - and more affordably - a simple halogen floodlight with an IR-pass filter over the front of it will work. A 30mm 50 watt "downlight" halogen would probably be adequate for this task; a cheap (to buy, if not to run...) 500 watt work-light flood, as beloved by skinflint digital photographers, will cover quite a large area. If you see a surveillance camera somewhere with one or more very-dull-red-glowing lights mounted next to it, you're looking at incandescent lights with IR-pass filters. The faint glow is because the filters can't quite block all of the visible light from the blazing filament in there.
The problem with filtered incandescents is that a proper professional filter could easily cost you more than the LED illuminator, and even then may burn up pretty quickly as it soaks up most of the few hundred watts of light the flood emits.
If you put the filter on a security light that has its own passive infra-red sensor that only turns the light on when the target is nearby, though, then it ought to work - and save you a considerable amount of electricity, too.
Realistically, this is only a solution I'd use if I already had a security light pointed the right way. But if you do (which indicates a certain audacity on the part of the Phantom Pool Filter Timer Resetter), then it could be worth trying.
I saw your blog entry mentioning Smokey Yunick, and I suddenly remembered reading about him and a revolutionary engine conversion he patented back then (must have been more than 15 years ago).
A couple of Swedish motor journalists had a test drive and they were floored by the power and torque you would get from a standard compact car with this thing mounted on it. Very efficient too, you barely needed a radiator, and you'd get crazy fuel efficiency as a bonus.
He also promised conversion kits for a reasonably low price. All in all it sounded pretty much like the stuff you usually debunk on a regular basis. :)
The thing was called "Hot Vapor", but sadly it never took off.
Was it too good to be true or was it for real?
The Hot Vapor "adiabatic engine" did indeed barely need a radiator, but the reason for that was that the engine ran very, very, very hot, by design.
I think this was the design's downfall. The thing had to run so hot that it'd burn itself, its oil, and the componentry surrounding it to death in quite short order.
That's no big deal for a drag car or similar short-endurance vehicle that gets torn down after every run, doesn't have lots of plastic, rubber, air conditioning gear etc in the engine bay, and has a driver who's wearing a Nomex suit. It's more of a problem for a road car. Even if you insulate the heck out of the engine, it'll still need very special oil.
It is, indeed, possible to make a heat engine much more efficient if you allow it to run very hot. It's right there in the basic equations. But I can believe that there's no way to turn a "hot vapor" engine into a workable hundred-thousand-mile proposition, at least not until we come up with some more advanced high temperature materials. And a bonnet latch that won't open until the engine's been stopped for half an hour.
I think the top post in this thread more or less covers it.
That guy has really lost the plot... or has he?
Clearly, selling a product with a couple of dollars worth of ingredients (if that) for $285.50 has to be very lucrative! His vague reference to Nikola Tesla would take some people in, I guess.
Honestly, I don't know what the state of mind of my old employer Tim at Megadisc is. I haven't been game to contact him.
Generally speaking, though, people who sell quacky products really do usually more or less believe in that stuff, insofar as they believe in anything.
(A lot of "New Age" beliefs are very hard to pin down. The more you talk to the people who claim to base their lives on them - or on "Wicca", or indeed on wishy-washy easygoing Christianity or Islam - the closer they get to admitting that all they've really got at the base of their belief system is "be excellent to each other, dudes". Which is OK with me; I wish more people were like that. But it's hardly a basis for believing that crystals/prayer/holy relics can cure cancer.)
I'm inclined to think that Megadisc-Tim is pretty sincere. People who know some goofy thing is crap and are selling it anyway usually have better looking Web sites than Tim's.
When I knew him, Tim was into kung fu (his washing machine sounded as if there were anvils in there). There's no essential connection between chop-socky and New Age balderdash, of course, but the two do often go together; I've heard more than once from people who've had trouble finding a place to learn martial arts that doesn't also teach metaphysical bulldust.
Buying for $3 and selling for $300 can indeed be the road to riches. But Tim's wholesale price may not actually be very much lower than the retail one. Don't assume that the guy on the end of the scam chain is the one who's raking it in.