Archive for June, 2008

Demo machine

Monday, June 30th, 2008

If one were to design a computer specifically for the purpose of having Oldskool-style demos written for it, what would it be like?

I came up with a few requirements:

  • Oldskool "feel"
  • Attractive output: sampled sound and realistic looking (still) images can be achieved very easily
  • The more skilled the programmer, the better effects can be achieved - very great skill is required for the best effects
  • Hardware as simple as possible
  • Easy to emulate
  • Fun to program

To get that Oldskool feel but still allow realistic looking still images, I decided to base it around a composite NTSC signal. It turns out there is a standard for digital composite NTSC signals - SMPTE 244M, so I just picked the 8-bit version of that. The video hardware is about as simple as video hardware could possibly be - just an 8-bit DAC that repeatedly cycles through a consecutive region of memory and dumps the bytes to the output at a frequency of 14.318MHz (4 times the color carrier frequency). This is also the CPU clock speed (the CPU is in sync with the pixel clock).

Technically, a still image in this format that is fully NTSC compliant requires 955,500 bytes (933Kb) of memory (2 frames*910*525) because the color carrier phase alternates between frames. Given that the machine's memory size should be a power of 2 (so that we can map 32-bit addresses to physical addresses just by clearing the top bits) I decided to give it 1Mb (any more and it stops being so Oldskool). This doesn't necessarily mean that you need to use almost all the memory to store the framebuffer - by tweaking the burst/sync patterns you can get a 912*525 (468Kb) interlaced mode where you only need one frame, or a 912*262 (233Kb) non-interlaced mode (again with only frame), both of which should be compatible with any composite monitor. These have usable resolutions (including the overscan area) of about 720*480 and 720*240 respectively (with respective CGA-esque pixel aspect ratios of 5:6 and 5:12). So this gives a nice bit of flexibility to software with no extra hardware complexity.

One disadvantage of this setup is that it may be possible for software to damage some composite monitors by creating sync pulses too often. So one would probably want to use an emulator when testing/debugging programs! Also, scrolling is fiddly because you have to move the image data independently of the sync/burst pulses. There are block move instructions which can move 4 pixels per clock to help with this, though.

Audio is done in a very similar way to video - set the beginning and end of the region of memory and the hardware cycles through that memory range turning bytes into samples and putting them through a DAC. Another register is used to set the sample rate (I decided a fixed rate would be too inflexible).

Programs are just 1Mb memory images which are dumped into RAM. The first two words of RAM are the instruction pointer and the stack pointer, so execution starts at the address pointed to by the word at 0. I/O registers are also at the start of the memory region.

Most instructions are 1 cycle long (with the exception of instructions like the equivalents of "REP MOVSB" and "REP STOSB") to enable effects that require cycle counting to be done as easily as possible. Most instructions are 1 byte long (with the exception of instructions that have a 1-byte, 2-byte or 4-byte immediate argument). The CPU is 32-bit (to make addressing that 1Mb of memory easy). The architecture is stack machine based (kind of like Java) - I have a soft spot for such architectures because they're really easy to write decent compilers for (registers are helpful at the hardware level for making processors fast, but that isn't a particular concern here). Devolving the CPU has some ideas for generating a good instruction set.

Because of the cycle counting and simplicity requirements, there are no hardware interrupts - all timing must be done by cycle counting. This also means that there are no instructions whose purpose is to determine some value N and take O(N) time to do it - so no "REP SCASB" in this architecture. I don't think that is very useful for demos anyway, and it's a non-goal of this design to be suitable for general purpose computing tasks like searching and parsing.

It would be pretty easy to generalize this architecture to make a games machine - just have a register whose bits reflect the up/down status of controller buttons.

Now, while I'd like to believe that building such a machine would revitalize the Oldskool demo scene by providing a new focal point, I suspect a more realistic point of view is that because programming such a machine would be such a non-applicable skill, because the limitations of the machine constrain the possible effects, and because there would be none of the nostalgia inspired by actual Oldskool hardware, nobody would actually be interested. But it's still an interesting exercise!


Sunday, June 29th, 2008

A fun toy would be a musical instrument with a microphone that can detect and modify pitch in real time. So if you sing an A but play a B it will transpose your voice up a tone. I understand such things are common in modern recording studios but I'm not sure if they work in real time (besides which, I don't think it would be too difficult to create such a thing from scratch).

Better still would be if it was polyphonic, so you sing a note, play a chord and hear a trio.

My favorite programming font

Saturday, June 28th, 2008

Many colleagues, on seeing my screen for the first time, have been horrified at how small the text on my screen is. I tend to use 1600x1200 monitors, and a 6x8 pixel font for coding purposes. I have good vision at normal head/screen distances and I like to be able to see a lot of text at once (I can see the "big picture" and also the details without having to explicitly zoom).

My preferred font is the 6 point Terminal font that comes with Windows. When I started my new job, I switched to a Linux machine for most of my day-to-day work and one thing that took me a long time to figure out was how to get my terminal windows to display this font. I got it working in XTerm but copy and paste works better in Gnome terminal, which refused to acknowledge any non-TrueType fonts or even bitmap fonts in a TrueType package.

I eventually went through the same route that this guy went through to create his fonts - I extracted the bitmaps from the .fon file and turned them into outlines by creating a square for each dark pixel. The resulting .ttf file is 30 times as big as the original bitmap but it seems to work fine with everything I've thrown at it.

Here is the resulting TrueType font and here is what it looks like:

Plug-ins should not be in the same address space as the main process

Friday, June 27th, 2008

After some thought I have come to the conclusion that a software architecture which allows third-party code to run in the same process as the main program is a bad idea.

The first problem with such an architecture is reliability - how do you protect yourself against bugs in the third-party code? The short answer is that you can't - the plugin has the ability to stomp all over your process's invariants. The best you can do is terminate the process when such a situation is detected.

The second problem is that a compiler can't reason about the third-party code (it might not even exist at compile time). This means that there are all kinds of checks and optimizations that cannot be performed (like some of the stack tricks I mentioned a while ago).

The third problem is that one cannot tell what code the plug-in will rely on - if it sticks to documented interfaces that's okay, but (either deliberately or accidentally) a plug-in might rely on undocumented behavior which causes the plug-in to break if the program is updated. This forces the program to have ugly kludges to keep old plug-ins working.

If plug-ins are run as separate processes communicating with the main program via well-defined IPC protocols, all these problems are solved. The down-side is that these protocols are likely to be a little more difficult to code against and (depending on the efficiency of the operating system's IPC implementation) may also be significantly slower. The speed problem seems unlikely to be insurmountable though - current OSes can play back HD video which involves the uncompressed video stream crossing the user/kernel boundary - few applications are likely to need IPC bandwidth greater than that.

Fine structure constant update

Thursday, June 26th, 2008

Many years ago I posted this on sci.physics. It turns out that the value of the Inverse Fine Structure Constant (a dimensionless parameter which can be experimentally measured as about 137.036 but for which no theory of physics yet predicts a value) is remarkably close to (alpha^2)(pi^2)(pi^pi-1)/16 where alpha is the second Feigenbaum constant, about 2.502907875096. This formula gives a value for the IFSC of 137.0359996810.

After posting that, I got a message from James Gilson pointing out his work on the same subject - he has a different formula for the IFSC, pi/(29*cos(pi/137)*tan(pi/(137*29))), which is not quite so pretty but does have the advantage of having some geometrical justification (which I still don't completely understand). Gilson's formula gives a value for the IFSC as 137.0359997867.

Back in 2001 the most accurate measurement of the IFSC (CODATA 1999) gave a value of 137.03599976(50) (i.e. there is a 68% chance that the true value is between 137.03599926 and 137.03600026). Both the formula give answers in this range.

I thought I would revisit this briefly and see if the latest measurements were able to rule out one of both of these formulae. Work of G. Gabrielse et al in 2006 give the IFSC as 137.035999068(96), i.e. there is a 68% chance that the true value is between 137.035998972 and 137.035999164. This appears to rule out both formulae. An earlier version of the 2006 Harvard work (which was compatible with both formulae) was superceded by an erratum. I admit this post is a bit anticlimactic but when I started writing it I thought that the latest measurements ruled out Gilson's formula but not mine.

I want to run as root

Wednesday, June 25th, 2008

With all the hype about Vista's User Access Control functionality, an important point seems to have gone largely unsaid. UAC protects the integrity of the operating system, but does nothing for the user's files. If I run some bad software in a non-user account, it can still read my financial documents, delete my photos etc. These things are much more difficult to fix than the problems UAC does prevent (which can be fixed by reinstalling the operating system).

The trend towards more Unix-like operating system structure annoys me somewhat. I want to run as root/admin all the time. If ask the for some critical system files, the operating system shouldn't second guess me, it should just do what I asked. I have been running Windows systems as administrator for years and it has never been a problem for me in practice. I don't ever want to have to input my password for machines that I own that I'm sitting in front of (remote access is different).

I think a better security model for a single-user machine would be not to authorize individual commands but to authorize programs. When a piece of software is downloaded from the internet and run, the OS calls it makes should be sandboxed. If it attempts to modify the system directories the OS should fake it so the system directories are not modified but so that it looks to that application like they have been. Private user files should just not appear to be present at all (not even findable but read-locked).

Vista is actually capable of this to some extent but it is only used as a last resort, to enable legacy programs to run. Applications released since Vista tend to have manifests which allow them to fail instead of get lied to - I don't think a program should even have the ability to tell that it is being lied to - if I want to lie to my media player and tell it that the sound output is going to a speaker when in fact it is going to a file, I should be able to do that. This is similar (but not quite the same) as a chroot jail in Unix, though chroot is not intended to be used as a sandbox for untrusted applications.

I suppose that, having said all that, what Microsoft have done in Vista does make sense for them - in the long run it will probably reduce support calls and promote development of software that requires only the minimum of privileges to run. I just wish I could turn it off completely without the annoying side-effects.

Map showing river basins

Tuesday, June 24th, 2008

I'd be interested to see a map of the UK where each point was coloured according to where the water that falls on the land corresponding to that point reaches the ocean.

Here's an example of the kind of thing I'm thinking of, except that this isn't very detailed - while it looks right for the Thames basin, lots of smaller rivers get bunched together (for example in the South-West).

This is a bit more like it but for Africa. Note that there are some regions which do not appear to be adjacent to any coast - this means that all the water from these areas ends up in interior lakes and ultimately evaporates again. They are Endorheic basins.

Shepard tones

Monday, June 23rd, 2008

If you arrange the musical notes C, C#, D, D#, E, F, F#, G, G#, A, A# around a circle (in that order), like this:

The resulting structure has some interesting properties. Firstly, each interval becomes a directed chord (in the geometric sense) on this circle, and transposition is just rotation.

The trouble is that because of octaves, this picture doesn't tell the whole story. If you go all the way around the circle you'll find yourself an octave higher (or lower) than when you started. Not all the intervals will sound alike as some of them will include an extra octave factor.

Unless you use Shepard tones.

Shepard tones have frequency components of the same pitch class in all octaves (tapering off at low and high frequencies according to a bell curve in log frequency space), like this:

They sound kind of like notes on a church organ.

The Shepard tones sound like musical notes but because the peak of the bell curve is always the same frequency, once you have progressed through a full scale you're back where you started instead of being up an octave. This can be used to create an auditory illusion of a pitch which sounds like it is continually getting higher (or lower), but is in fact looping.

It would be interesting to compose some music using Shepard tones. Another interesting feature of them is that all chords are identical to their own inversions. So in a number of ways Shepard tones drastically simplify music theory.

Another fun thing to do with Shepard tones is to leave the pitch class alone and vary the frequency that is at the peak of the bell curve. That way you can make a sound which continuously increases in frequency but which never seems to change pitch.

Widdershins of hue

Sunday, June 22nd, 2008

Here is a concept that I feel that there ought to be words for but which there don't seem to be.

Suppose we arrange the spectrum of colours into the familiar colour wheel:

What is the word that corresponds to "clockwise" on this diagram (and "anticlockwise" on the mirror image of this diagram)? I.e. the word that means "more blue than red", "more green than blue" and "more red than green" simultaneously? What is the word for the opposite hue angle direction?

Bonus points for the best neologisms for these equal and opposite concepts.

On "Taking Science on Faith"

Saturday, June 21st, 2008

This article caused quite a stir and many responses when it appeared, but none of the responses I read spotted the gaping hole in Davies' reasoning.

No faith is necessary in Physics - just measurement. The game is this - we come up with a theory that correctly predict the results of all experiments done so far, from that we determine an experiment that we haven't done yet that would confirm or disprove this theory, and we perform the experiment. If we disprove it, then we we come up with another theory and repeat. If we confirm it we try to think of another experiment. Once we have a single, consistent theory that explains all experimental results, we're done (okay, there's still the small matter of determining all the consequences of that theory but fundamental physics is done).

If the laws of physics are found to "vary from place to place on a mega-cosmic scale" that doesn't make the game impossible to play, it's just another experimental result to be slotted into the next generation of theories. We'd discover the "meta rules" that governed how the rules of physics changed from place to place. And maybe there are meta-meta-rules that govern how the meta-rules change and so on.

This tower might never end, of course - a theory describing all experimental observations might take an infinite amount of information to describe. That's not a problem - it just means that the game never ends. Historically, new theories of physics have begotten new technologies, so as we discover more of this tower it stands to reason that our technologies would become more and more powerful.

The alternative is that the tower does end - there is a final theory of physics. This alternative seems more likely to me - historically our theories of physics have actually become simpler (for the amount of stuff they explain) over time. Together just two theories (General Relativity and the Standard Model of Quantum Mechanics) can account for the results of all experiments we have been able to do so far. These theories are simple enough that they can both be learnt over the course of a four-year undergraduate university degree.

I suspect that the theory that finally unifies these will be simpler still if you know how to look at it in the right way. In fact, I think that it will be so elegant and simple to state that it will be unquestionably correct - the reaction of a physicist from today on seeing and understanding it would be "Of course! How could it possibly be any other way!"

The only "faith" that physicists need to have is the faith that we can gain some knowledge about things by observing them (performing experiments) - which seems to me to be more of a tautology than a matter of faith.

Davies suggests an alternative, which is "to regard the laws of physics and the universe they govern as part and parcel of a unitary system, and to be incorporated together within a common explanatory scheme. In other words, the laws should have an explanation from within the universe and not involve appealing to an external agency." I'm not even sure what that means and I don't think Davies does either. I can't bring myself to believe that there is some box in a dusty attic in some corner of a distant galaxy which somehow "explains the laws of physics".