Related
I am writing a program in C++ to reconstruct a 3D object from a set of projected 2D images, the most computation-intensive part of which involves magnifying and shifting each image via bilinear interpolation. I currently have a pair of functions for this task; "blnSetup" defines a handful of parameters outside the loop, then "bilinear" applies the interpolation point-by-point within the loop:
(NOTE: 'I' is a 1D array containing ordered rows of image data)
//Pre-definition structure (in header)
struct blnData{
float* X;
float* Y;
int* I;
float X0;
float Y0;
float delX;
float delY;
};
//Pre-definition function (outside the FOR loop)
extern inline blnData blnSetup(float* X, float* Y, int* I)
{
blnData bln;
//Create pointers to X, Y, and I vectors
bln.X = X;
bln.Y = Y;
bln.I = I;
//Store offset and step values for X and Y
bln.X0 = X[0];
bln.delX = X[1] - X[0];
bln.Y0 = Y[0];
bln.delY = Y[1] - Y[0];
return bln;
}
//Main interpolation function (inside the FOR loop)
extern inline float bilinear(float x, float y, blnData bln)
{
float Ixy;
//Return -1 if the target point is outside the image matrix
if (x < bln.X[0] || x > bln.X[-1] || y < bln.Y[0] || y > bln.Y[-1])
Ixy = 0;
//Otherwise, apply bilinear interpolation
else
{
//Define known image width
int W = 200;
//Find nearest indices for interpolation
int i = floor((x - bln.X0) / bln.delX);
int j = floor((y - bln.Y0) / bln.delY);
//Interpolate I at (xi, yj)
Ixy = 1 / ((bln.X[i + 1] - bln.X[i])*(bln.Y[j + 1] - bln.Y[j])) *
(
bln.I[W*j + i] * (bln.X[i + 1] - x) * (bln.Y[j + 1] - y) +
bln.I[W*j + i + 1] * (x - bln.X[i]) * (bln.Y[j + 1] - y) +
bln.I[W*(j + 1) + i] * (bln.X[i + 1] - x) * (y - bln.Y[j]) +
bln.I[W*(j + 1) + i + 1] * (x - bln.X[i]) * (y - bln.Y[j])
);
}
return Ixy;
}
EDIT: The function calls are below. 'flat.imgdata' is a std::vector containing the input image data and 'proj.imgdata' is a std::vector containing the transformed image.
int Xs = flat.dim[0];
int Ys = flat.dim[1];
int* Iarr = flat.imgdata.data();
float II, x, y;
bln = blnSetup(X, Y, Iarr);
for (int j = 0; j < flat.imgdata.size(); j++)
{
x = 1.2*X[j % Xs];
y = 1.2*Y[j / Xs];
II = bilinear(x, y, bln);
proj.imgdata[j] = (int)II;
}
Since I started optimizing, I have been able to reduce computation time by ~50x (!) by switching from std::vectors to C arrays within the interpolation function, and another 2x or so by cleaning up redundant computations/typecasting/etc, but assuming O(n) with n being the total number of processed pixels, the full reconstruction (~7e10 pixels) should still take 40min or so--about an order of magnitude longer than my goal of <5min.
According to Visual Studio's performance profiler, the interpolation function call ("II = bilinear(x, y, bln);") is unsurprisingly still the majority of my computation load. I haven't been able to find any linear algebraic methods for fast multiple interpolation, so my question is: is this basically as fast as my code will get, short of applying more or faster CPUs to the task? Or is there a different approach that might speed things up?
P.S. I've also only been coding in C++ for about a month now, so feel free to point out any beginner mistakes I might be making.
I wrote up a long answer suggesting looking at OpenCV (opencv.org), or using Halide (http://halide-lang.org/), and getting into how image warping is optimized, but I think a shorter answer might serve better. If you are really just scaling and translating entire images, OpenCV has code to do that and we have an example for resizing in Halide as well (https://github.com/halide/Halide/blob/master/apps/resize/resize.cpp).
If you really have an algorithm that needs to index an image using floating-point coordinates which result from a computation that cannot be turned into a moderately simple function on integer coordinates, then you really want to be using filtered texture sampling on a GPU. Most techniques for optimizing on the CPU rely on exploiting some regular pattern of access in the algorithm and removing float to integer conversion from the addressing. (For resizing, one uses two integer variables, one which indexes the pixel coordinate of the image and the other which is the fractional part of the coordinate and it indexes a kernel of weights.) If this is not possible, the speedups are somewhat limited on CPUs. OpenCV does provide fairly general remapping support, but it likely isn't all that fast.
Two optimizations that may be applicable here are trying to move the boundary condition out the loop and using a two pass approach in which the horizontal and vertical dimensions are processed separately. The latter may or may not win and will require tiling the data to fit in cache if the images are very large. Tiling in general is pretty important for large images, but it isn't clear it is the first order performance problem here and depending on the values in the inputs, the cache behavior may not be regular enough anyway.
"vector 50x slower than array". That's a dead giveaway you're in debug mode, where vector::operator[] is not inlined. You will probably get the necessary speedup, and a lot more, simply by switching to release mode.
As a bonus, vector has a .back() method, so you have a proper replacement for that [-1]. Pointers to the begin of an array don't contain the array size, so you can't find the back of an array that way.
I'm using bits of Perry Cook's Synthesis Toolkit (STK) to generate saw and square waves. STK includes this BLIT-based sawtooth oscillator:
inline STKFloat BlitSaw::tick( void ) {
StkFloat tmp, denominator = sin( phase_ );
if ( fabs(denominator) <= std::numeric_limits<StkFloat>::epsilon() )
tmp = a_;
else {
tmp = sin( m_ * phase_ );
tmp /= p_ * denominator;
}
tmp += state_ - C2_;
state_ = tmp * 0.995;
phase_ += rate_;
if ( phase_ >= PI )
phase_ -= PI;
lastFrame_[0] = tmp;
return lastFrame_[0];
}
The square wave oscillator is broadly similar. At the top, there's this comment:
// A fully optimized version of this code would replace the two sin
// calls with a pair of fast sin oscillators, for which stable fast
// two-multiply algorithms are well known.
I don't know where to start looking for these "fast two-multiply algorithms" and I'd appreciate some pointers. I could use a lookup table instead, but I'm keen to learn what these 'fast sin oscillators' are. I could also use an abbreviated Taylor series, but thats way more than two multiplies. Searching hasn't turned up anything much, although I did find this approximation:
#define AD_SIN(n) (n*(2.f- fabs(n)))
Plotting it out shows that it's not really a close approximation outside the range of -1 to 1, so I don't think I can use it when phase_ is in the range -pi to pi:
Here, Sine is the blue line and the purple line is the approximation.
Profiling my code reveals that the calls to sin() are far and away the most time-consuming calls, so I really would like to optimise this piece.
Thanks
EDIT Thanks for the detailed and varied answers. I will explore these and accept one at the weekend.
EDIT 2 Would the anonymous close voter please kindly explain their vote in the comments? Thank you.
Essentially the sinusoidal oscilator is one (or more) variables that change with each DSP step, rather than getting recalculated from scratch.
The simplest are based on the following trig identities: (where d is constant, and thus so is cos(d) and sin(d) )
sin(x+d) = sin(x) cos(d) + cos(x) sin(d)
cos(x+d) = cos(x) cos(d) - sin(x) sin(d)
However this requires two variables (one for sin and one for cos) and 4 multiplications to update. However this will still be far faster than calculating a full sine at each step.
The solution by Oli Charlesworth is based on solutions to this general equation
A_{n+1} = a A_{n} + A_{n-1}
Where looking for a solution of the form A_n = k e^(i theta n) gives an equation for theta.
e^(i theta (n+1) ) = a e^(i theta n ) + b e^(i theta (n-1) )
Which simplifies to
e^(i theta) - e^(-i theta ) = a
2 cos(theta) = a
Giving
A_{n+1} = 2 cos(theta) A_{n} + A_{n-1}
Whichever approach you use you'll either need to use one or two of these oscillators for each frequency, or use another trig identity to derive the higher or lower frequencies.
How accurate do you need this?
This function, f(x)=0.398x*(3.1076-|x|), does a reasonably good job for x between -pi and pi.
Edit
An even better approximation is f(x)=0.38981969947653056*(pi-|x|), which keeps the absolute error to 0.038158444604 or less for x between -pi and pi.
A least squares minimization will yield a slightly different function.
It's not possible to generate one-off sin calls with just two multiplies (well, not a useful approximation, at any rate). But it is possible to generate an oscillator with low complexity, i.e. where each value is calculated in terms of the preceding ones.
For instance, consider that the following difference equation will give you a sinusoid:
y[n] = 2*cos(phi)*y[n-1] - y[n-2]
(where cos(phi) is a constant)
(From the original author of the VST BLT code).
As a matter of fact, I was porting the VST BLT oscillators to C#, so I was googling for good sin oscillators. Here's what I came up with. Translation to C++ is straightforward. See the notes at the end about accuumulated round-off errors.
public class FastOscillator
{
private double b1;
private double y1, y2;
private double fScale;
public void Initialize(int sampleRate)
{
fScale = AudioMath.TwoPi / sampleRate;
}
// frequency in Hz. phase in radians.
public void Start(float frequency, double phase)
{
double w = frequency * fScale;
b1 = 2.0 * Math.Cos(w);
y1 = Math.Sin(phase - w);
y2 = Math.Sin(phase - w * 2);
}
public double Tick()
{
double y0 = b1 * y1 - y2;
y2 = y1;
y1 = y0;
return y0;
}
}
Note that this particular oscillator implementation will drift over time, so it needs to be re-initialzed periodically. In this particular implementation, the magnitude of the sin wave decays over time. The original comments in the STK code suggested a two-multiply oscillator. There are, in fact, two-multiply oscillators that are reasonably stable over time. But in retrospect, the need to keep the sin(phase), and sin(m*phase) oscillators tightly in synch probably means that they have to be resynched anyway. Round-off errors between phase and m*phase mean that even if the oscillators were stable, they would drift eventually, running a significant risk of producing large spikes in values near the zeros of the BLT functions. May as well use a one-multiply oscillator.
These particular oscillators should probably be re-initialized every 30 to 100 cycles (or so). My C# implementation is frame based (i.e. it calculates an float[] array of results in a void Tick(int count, float[] result) method. The oscillators are re-synched at the end of each Tick call. Something like this:
void Tick(int count, float[] result)
{
for (int i = 0; i < count; ++i)
{
...
result[i] = bltResult;
}
// re-initialize the oscillators to avoid accumulated drift.
this.phase = (this.phase + this.dPhase*count) % AudioMath.TwoPi;
this.sinOsc.Initialize(frequency,this.phase);
this.mSinOsc.Initialize(frequency*m,this.phase*m);
}
Probably missing from the STK code. You might want to investigate this. The original code provided to the STK did this. Gary Scavone tweaked the code a bit, and I think the optimization was lost. I do know that the STK implementations suffer from DC drift, which can be almost entirely eliminated when implemented properly.
There's a peculiar hack that prevents DC drift of the oscillators, even when sweeping the frequency of the oscillators. The trick is that the oscillators should be started with an initial phase adjustment of dPhase/2. That just so happens to start the oscillators off with zero DC drift, without having to figure out wat the correct initial state for various integrators in each of the BLT oscillators.
Strangely, if the adjustment is re-adjusted whenever the frequency of the oscillator changes, then this also prevents wild DC drift of the output when sweeping the frequency of the oscillator. Whenever the frequency changes, subtract dPhase/2 from the previous phase value, recalculate dPhase for the new frequency, and then add dPhase/2.I rather suspect this could be formally proven; but I have not been able to so. All I know is that It Just Works.
For a block implementation, the oscillators should actually be initialized as follows, instead of carrying the phase adjustment in the current this.phase value.
this.sinOsc.Initialize(frequency,phase+dPhase*0.5);
this.mSinOsc.Initialize(frequency*m,(phase+dPhase*0.5)*m);
You might want to take a look here:
http://devmaster.net/forums/topic/4648-fast-and-accurate-sinecosine/
There's some sample code that calculates a very good appoximation of sin/cos using only multiplies, additions and the abs() function. Quite fast too. The comments are also a good read.
It essentiall boils down to this:
float sine(float x)
{
const float B = 4/pi;
const float C = -4/(pi*pi);
const float P = 0.225;
float y = B * x + C * x * abs(x);
return P * (y * abs(y) - y) + y;
}
and works for a range of -PI to PI
If you can, you should consider memorization based techniques. Essentially store sin(x) and cos(x) values for a bunch values. To calculate sin(y), find a and b for which precomputed values exist such that a<=y<=b. Now using sin(a), sin(b), cos(a), cos(b), y-a and y-b approximately calculate sin(y).
The general idea of getting periodically sampled results from the sine or cosine function is to use a trig recursion or an initialized (barely) stable IIR filter (which can end up being pretty much the same computations). There are bunches of these in the DSP literature, of varying accuracy and stability. Choose carefully.
My game needs to move by a certain angle. To do this I get the vector of the angle via sin and cos. Unfortunately sin and cos are my bottleneck. I'm sure I do not need this much precision. Is there an alternative to a C sin & cos and look-up table that is decently precise but very fast?
I had found this:
float Skeleton::fastSin( float x )
{
const float B = 4.0f/pi;
const float C = -4.0f/(pi*pi);
float y = B * x + C * x * abs(x);
const float P = 0.225f;
return P * (y * abs(y) - y) + y;
}
Unfortunately, this does not seem to work. I get significantly different behavior when I use this sin rather than C sin.
Thanks
A lookup table is the standard solution. You could Also use two lookup tables on for degrees and one for tenths of degrees and utilize sin(A + B) = sin(a)cos(b) + cos(A)sin(b)
For your fastSin(), you should check its documentation to see what range it's valid on. The units you're using for your game could be too big or too small and scaling them to fit within that function's expected range could make it work better.
EDIT:
Someone else mentioned getting it into the desired range by subtracting PI, but apparently there's a function called fmod for doing modulus division on floats/doubles, so this should do it:
#include <iostream>
#include <cmath>
float fastSin( float x ){
x = fmod(x + M_PI, M_PI * 2) - M_PI; // restrict x so that -M_PI < x < M_PI
const float B = 4.0f/M_PI;
const float C = -4.0f/(M_PI*M_PI);
float y = B * x + C * x * std::abs(x);
const float P = 0.225f;
return P * (y * std::abs(y) - y) + y;
}
int main() {
std::cout << fastSin(100.0) << '\n' << std::sin(100.0) << std::endl;
}
I have no idea how expensive fmod is though, so I'm going to try a quick benchmark next.
Benchmark Results
I compiled this with -O2 and ran the result with the Unix time program:
int main() {
float a = 0;
for(int i = 0; i < REPETITIONS; i++) {
a += sin(i); // or fastSin(i);
}
std::cout << a << std::endl;
}
The result is that sin is about 1.8x slower (if fastSin takes 5 seconds, sin takes 9). The accuracy also seemed to be pretty good.
If you chose to go this route, make sure to compile with optimization on (-O2 in gcc).
I know this is already an old topic, but for people who have the same question, here is a tip.
A lot of times in 2D and 3D rotation, all vectors are rotated with a fixed angle. In stead of calling the cos() or sin() every cycle of the loop, create variable before the loop which contains the value of cos(angle) or sin(angle) already. You can use this variable in your loop. This way the function only has to be called once.
If you rephrase the return in fastSin as
return (1-P) * y + P * (y * abs(y))
And rewrite y as (for x>0 )
y = 4 * x * (pi-x) / (pi * pi)
you can see that y is a parabolic first-order approximation to sin(x) chosen so that it passes through (0,0), (pi/2,1) and (pi,0), and is symmetrical about x=pi/2.
Thus we can only expect our function to be a good approximation from 0 to pi. If we want values outside that range we can use the 2-pi periodicity of sin(x) and that sin(x+pi) = -sin(x).
The y*abs(y) is a "correction term" which also passes through those three points. (I'm not sure why y*abs(y) is used rather than just y*y since y is positive in the 0-pi range).
This form of overall approximation function guarantees that a linear blend of the two functions y and y*y, (1-P)*y + P * y*y will also pass through (0,0), (pi/2,1) and (pi,0).
We might expect y to be a decent approximation to sin(x), but the hope is that by picking a good value for P we get a better approximation.
One question is "How was P chosen?". Personally, I'd chose the P that produced the least RMS error over the 0,pi/2 interval. (I'm not sure that's how this P was chosen though)
Minimizing this wrt. P gives
This can be rearranged and solved for p
Wolfram alpha evaluates the initial integral to be the quadratic
E = (16 π^5 p^2 - (96 π^5 + 100800 π^2 - 967680)p + 651 π^5 - 20160 π^2)/(1260 π^4)
which has a minimum of
min(E) = -11612160/π^9 + 2419200/π^7 - 126000/π^5 - 2304/π^4 + 224/π^2 + (169 π)/420
≈ 5.582129689596371e-07
at
p = 3 + 30240/π^5 - 3150/π^3
≈ 0.2248391013559825
Which is pretty close to the specified P=0.225.
You can raise the accuracy of the approximation by adding an additional correction term. giving a form something like return (1-a-b)*y + a y * abs(y) + b y * y * abs(y). I would find a and b by in the same way as above, this time giving a system of two linear equations in a and b to solve, rather than a single equation in p. I'm not going to do the derivation as it is tedious and the conversion to latex images is painful... ;)
NOTE: When answering another question I thought of another valid choice for P.
The problem is that using reflection to extend the curve into (-pi,0) leaves a kink in the curve at x=0. However, I suspect we can choose P such that the kink becomes smooth.
To do this take the left and right derivatives at x=0 and ensure they are equal. This gives an equation for P.
You can compute a table S of 256 values, from sin(0) to sin(2 * pi). Then, to pick sin(x), bring back x in [0, 2 * pi], you can pick 2 values S[a], S[b] from the table, such as a < x < b. From this, linear interpolation, and you should have a fair approximation
memory saving trick : you actually need to store only from [0, pi / 2], and use symmetries of sin(x)
enhancement trick : linear interpolation can be a problem because of non-smooth derivatives, humans eyes is good at spotting such glitches in animation and graphics. Use cubic interpolation then.
What about
x*(0.0174532925199433-8.650935142277599*10^-7*x^2)
for deg and
x*(1-0.162716259904269*x^2)
for rad on -45, 45 and -pi/4 , pi/4 respectively?
This (i.e. the fastsin function) is approximating the sine function using a parabola. I suspect it's only good for values between -π and +π. Fortunately, you can keep adding or subtracting 2π until you get into this range. (Edited to specify what is approximating the sine function using a parabola.)
you can use this aproximation.
this solution use a quadratic curve :
http://www.starming.com/index.php?action=plugin&v=wave&ajax=iframe&iframe=fullviewonepost&mid=56&tid=4825
Is there a way to do a quick and dirty 3D distance check where the results are rough, but it is very very fast? I need to do depth sorting. I use STL sort like this:
bool sortfunc(CBox* a, CBox* b)
{
return a->Get3dDistance(Player.center,a->center) <
b->Get3dDistance(Player.center,b->center);
}
float CBox::Get3dDistance( Vec3 c1, Vec3 c2 )
{
//(Dx*Dx+Dy*Dy+Dz*Dz)^.5
float dx = c2.x - c1.x;
float dy = c2.y - c1.y;
float dz = c2.z - c1.z;
return sqrt((float)(dx * dx + dy * dy + dz * dz));
}
Is there possibly a way to do it without a square root or possibly without multiplication?
You can leave out the square root because for all positive (or really, non-negative) numbers x and y, if sqrt(x) < sqrt(y) then x < y. Since you're summing squares of real numbers, the square of every real number is non-negative, and the sum of any positive numbers is positive, the square root condition holds.
You cannot eliminate the multiplication, however, without changing the algorithm. Here's a counterexample: if x is (3, 1, 1) and y is (4, 0, 0), |x| < |y| because sqrt(1*1+1*1+3*3) < sqrt(4*4+0*0+0*0) and 1*1+1*1+3*3 < 4*4+0*0+0*0, but 1+1+3 > 4+0+0.
Since modern CPUs can compute a dot product faster than they can actually load the operands from memory, it's unlikely that you would have anything to gain by eliminating the multiply anyway (I think the newest CPUs have a special instruction that can compute a dot product every 3 cycles!).
I would not consider changing the algorithm without doing some profiling first. Your choice of algorithm will heavily depend on the size of your dataset (does it fit in cache?), how often you have to run it, and what you do with the results (collision detection? proximity? occlusion?).
What I usually do is first filter by Manhattan distance
float CBox::Within3DManhattanDistance( Vec3 c1, Vec3 c2, float distance )
{
float dx = abs(c2.x - c1.x);
float dy = abs(c2.y - c1.y);
float dz = abs(c2.z - c1.z);
if (dx > distance) return 0; // too far in x direction
if (dy > distance) return 0; // too far in y direction
if (dz > distance) return 0; // too far in z direction
return 1; // we're within the cube
}
Actually you can optimize this further if you know more about your environment. For example, in an environment where there is a ground like a flight simulator or a first person shooter game, the horizontal axis is very much larger than the vertical axis. In such an environment, if two objects are far apart they are very likely separated more by the x and y axis rather than the z axis (in a first person shooter most objects share the same z axis). So if you first compare x and y you can return early from the function and avoid doing extra calculations:
float CBox::Within3DManhattanDistance( Vec3 c1, Vec3 c2, float distance )
{
float dx = abs(c2.x - c1.x);
if (dx > distance) return 0; // too far in x direction
float dy = abs(c2.y - c1.y);
if (dy > distance) return 0; // too far in y direction
// since x and y distance are likely to be larger than
// z distance most of the time we don't need to execute
// the code below:
float dz = abs(c2.z - c1.z);
if (dz > distance) return 0; // too far in z direction
return 1; // we're within the cube
}
Sorry, I didn't realize the function is used for sorting. You can still use Manhattan distance to get a very rough first sort:
float CBox::ManhattanDistance( Vec3 c1, Vec3 c2 )
{
float dx = abs(c2.x - c1.x);
float dy = abs(c2.y - c1.y);
float dz = abs(c2.z - c1.z);
return dx+dy+dz;
}
After the rough first sort you can then take the topmost results, say the top 10 closest players, and re-sort using proper distance calculations.
Here's an equation that might help you get rid of both sqrt and multiply:
max(|dx|, |dy|, |dz|) <= distance(dx,dy,dz) <= |dx| + |dy| + |dz|
This gets you a range estimate for the distance which pins it down to within a factor of 3 (the upper and lower bounds can differ by at most 3x). You can then sort on, say, the lower number. You then need to process the array until you reach an object which is 3x farther away than the first obscuring object. You are then guaranteed to not find any object that is closer later in the array.
By the way, sorting is overkill here. A more efficient way would be to make a series of buckets with different distance estimates, say [1-3], [3-9], [9-27], .... Then put each element in a bucket. Process the buckets from smallest to largest until you reach an obscuring object. Process 1 additional bucket just to be sure.
By the way, floating point multiply is pretty fast nowadays. I'm not sure you gain much by converting it to absolute value.
I'm disappointed that the great old mathematical tricks seem to be getting lost. Here is the answer you're asking for. Source is Paul Hsieh's excellent web site: http://www.azillionmonkeys.com/qed/sqroot.html . Note that you don't care about distance; you will do fine for your sort with square of distance, which will be much faster.
In 2D, we can get a crude approximation of the distance metric without a square root with the formula:
distanceapprox (x, y) =
which will deviate from the true answer by at most about 8%. A similar derivation for 3 dimensions leads to:
distanceapprox (x, y, z) =
with a maximum error of about 16%.
However, something that should be pointed out, is that often the distance is only required for comparison purposes. For example, in the classical mandelbrot set (z←z2+c) calculation, the magnitude of a complex number is typically compared to a boundary radius length of 2. In these cases, one can simply drop the square root, by essentially squaring both sides of the comparison (since distances are always non-negative). That is to say:
√(Δx2+Δy2) < d is equivalent to Δx2+Δy2 < d2, if d ≥ 0
I should also mention that Chapter 13.2 of Richard G. Lyons's "Understanding Digital Signal Processing" has an incredible collection of 2D distance algorithms (a.k.a complex number magnitude approximations). As one example:
Max = x > y ? x : y;
Min = x < y ? x : y;
if ( Min < 0.04142135Max )
|V| = 0.99 * Max + 0.197 * Min;
else
|V| = 0.84 * Max + 0.561 * Min;
which has a maximum error of 1.0% from the actual distance. The penalty of course is that you're doing a couple branches; but even the "most accepted" answer to this question has at least three branches in it.
If you're serious about doing a super fast distance estimate to a specific precision, you could do so by writing your own simplified fsqrt() estimate using the same basic method as the compiler vendors do, but at a lower precision, by doing a fixed number of iterations. For example, you can eliminate the special case handling for extremely small or large numbers, and/or also reduce the number of Newton-Rapheson iterations. This was the key strategy underlying the so-called "Quake 3" fast inverse square root implementation -- it's the classic Newton algorithm with exactly one iteration.
Do not assume that your fsqrt() implementation is slow without benchmarking it and/or reading the sources. Most modern fsqrt() library implementations are branchless and really damned fast. Here for example is an old IBM floating point fsqrt implementation. Premature optimization is, and always will be, the root of all evil.
Note that for 2 (non-negative) distances A and B, if sqrt(A) < sqrt(B), then A < B. Create a specialized version of Get3DDistance() (GetSqrOf3DDistance()) that does not call sqrt() that would be used only for the sortfunc().
If you worry about performance, you should also take care of the way you send your arguments:
float Get3dDistance( Vec3 c1, Vec3 c2 );
implies two copies of Vec3 structure. Use references instead:
float Get3dDistance( Vec3 const & c1, Vec3 const & c2 );
You could compare squares of distances instead of the actual distances, since d2 = (x1-x2)2 + (y1-y2)2+ (z1-z2)2. It doesn't get rid of the multiplication, but it does eliminate the square root operation.
How often are the input vectors updated and how often are they sorted? Depending on your design, it might be quite efficient to extend the "Vec3" class with a pre-calculated distance and sort on that instead. Especially relevant if your implementation allows you to use vectorized operations.
Other than that, see the flipcode.com article on approximating distance functions for a discussion on yet another approach.
Depending slightly on the number of points that you are being used to compare with, what is below is pretty much guaranteed to be the get the list of points in approximate order assuming all points change at all iteration.
1) Rewrite the array into a single list of Manhattan distances with
out[ i ] = abs( posn[ i ].x - player.x ) + abs( posn[ i ].y - player.y ) + abs( posn[ i ].z - player.z );
2) Now you can use radix sort on floating point numbers to order them.
Note that in practice this is going to be a lot faster than sorting the list of 3d positions because it significantly reduces the memory bandwidth requirements in the sort operation which all of the time is going to be spend and in which unpredictable accesses and writes are going to occur. This will run on O(N) time.
If many of the points are stationary at each direction there are far faster algorithms like using KD-Trees, although implementation is quite a bit more complex and it is much harder to get good memory access patterns.
If this is simply a value for sorting, then you can swap the sqrt() for a abs(). If you need to compare distances against set values, get the square of that value.
E.g. instead of checking sqrt(...) against a, you can compare abs(...) against a*a.
You may want to consider caching the distance between the player and the object as you calculate it, and then use that in your sortfunc. This would depend upon how many times your sort function looks at each object, so you might have to profile to be sure.
What I'm getting at is that your sort function might do something like this:
compare(a,b);
compare(a,c);
compare(a,d);
and you would calculate the distance between the player and 'a' every time.
As others have mentioned, you can leave out the sqrt in this case.
If you could center your coordinates around the player, use spherical coordinates? Then you could sort by the radius.
That's a big if, though.
If your operation happens a lot, it might be worth to put it into some 3D data structure. You probably need the distance sorting to decide which object is visible, or some similar task. In order of complexity you can use:
Uniform (cubic) subdivision
Divide the used space into cells, and assign the objects to the cells. Fast access to element, neighbours are trivial, but empty cells take up a lot of space.
Quadtree
Given a threshold, divide used space recursively into four quads until less then threshold number of object is inside. Logarithmic access element if objects don't stack upon each other, neighbours are not hard to find, space efficient solution.
Octree
Same as Quadtree, but divides into 8, optimal even if objects are above each other.
Kd tree
Given some heuristic cost function, and a threshold, split space into two halves with a plane where the cost function is minimal. (Eg.: same amount of objects at each side.) Repeat recursively until threshold reached. Always logarithmic, neighbours are harder to get, space efficient (and works in all dimensions).
Using any of the above data structures, you can start from a position, and go from neighbour to neighbour to list the objects in increasing distance. You can stop at desired cut distance. You can also skip cells that cannot be seen from the camera.
For the distance check, you can do one of the above mentioned routines, but ultimately they wont scale well with increasing number of objects. These can be used to display data that takes hundreds of gigabytes of hard disc space.
How can I rewrite the following pseudocode in C++?
real array sine_table[-1000..1000]
for x from -1000 to 1000
sine_table[x] := sine(pi * x / 1000)
I need to create a sine_table lookup table.
You can reduce the size of your table to 25% of the original by only storing values for the first quadrant, i.e. for x in [0,pi/2].
To do that your lookup routine just needs to map all values of x to the first quadrant using simple trig identities:
sin(x) = - sin(-x), to map from quadrant IV to I
sin(x) = sin(pi - x), to map from quadrant II to I
To map from quadrant III to I, apply both identities, i.e. sin(x) = - sin (pi + x)
Whether this strategy helps depends on how much memory usage matters in your case. But it seems wasteful to store four times as many values as you need just to avoid a comparison and subtraction or two during lookup.
I second Jeremy's recommendation to measure whether building a table is better than just using std::sin(). Even with the original large table, you'll have to spend cycles during each table lookup to convert the argument to the closest increment of pi/1000, and you'll lose some accuracy in the process.
If you're really trying to trade accuracy for speed, you might try approximating the sin() function using just the first few terms of the Taylor series expansion.
sin(x) = x - x^3/3! + x^5/5! ..., where ^ represents raising to a power and ! represents the factorial.
Of course, for efficiency, you should precompute the factorials and make use of the lower powers of x to compute higher ones, e.g. use x^3 when computing x^5.
One final point, the truncated Taylor series above is more accurate for values closer to zero, so its still worthwhile to map to the first or fourth quadrant before computing the approximate sine.
Addendum:
Yet one more potential improvement based on two observations:
1. You can compute any trig function if you can compute both the sine and cosine in the first octant [0,pi/4]
2. The Taylor series expansion centered at zero is more accurate near zero
So if you decide to use a truncated Taylor series, then you can improve accuracy (or use fewer terms for similar accuracy) by mapping to either the sine or cosine to get the angle in the range [0,pi/4] using identities like sin(x) = cos(pi/2-x) and cos(x) = sin(pi/2-x) in addition to the ones above (for example, if x > pi/4 once you've mapped to the first quadrant.)
Or if you decide to use a table lookup for both the sine and cosine, you could get by with two smaller tables that only covered the range [0,pi/4] at the expense of another possible comparison and subtraction on lookup to map to the smaller range. Then you could either use less memory for the tables, or use the same memory but provide finer granularity and accuracy.
long double sine_table[2001];
for (int index = 0; index < 2001; index++)
{
sine_table[index] = std::sin(PI * (index - 1000) / 1000.0);
}
One more point: calling trigonometric functions is pricey. if you want to prepare the lookup table for sine with constant step - you may save the calculation time, in expense of some potential precision loss.
Consider your minimal step is "a". That is, you need sin(a), sin(2a), sin(3a), ...
Then you may do the following trick: First calculate sin(a) and cos(a). Then for every consecutive step use the following trigonometric equalities:
sin([n+1] * a) = sin(n*a) * cos(a) + cos(n*a) * sin(a)
cos([n+1] * a) = cos(n*a) * cos(a) - sin(n*a) * sin(a)
The drawback of this method is that during this procedure the round-off error is accumulated.
double table[1000] = {0};
for (int i = 1; i <= 1000; i++)
{
sine_table[i-1] = std::sin(PI * i/ 1000.0);
}
double getSineValue(int multipleOfPi){
if(multipleOfPi == 0) return 0.0;
int sign = 1;
if(multipleOfPi < 0){
sign = -1;
}
return signsine_table[signmultipleOfPi - 1];
}
You can reduce the array length to 500, by a trick sin(pi/2 +/- angle) = +/- cos(angle).
So store sin and cos from 0 to pi/4.
I don't remember from top of my head but it increased the speed of my program.
You'll want the std::sin() function from <cmath>.
another approximation from a book or something
streamin ramp;
streamout sine;
float x,rect,k,i,j;
x = ramp -0.5;
rect = x * (1 - x < 0 & 2);
k = (rect + 0.42493299) *(rect -0.5) * (rect - 0.92493302) ;
i = 0.436501 + (rect * (rect + 1.05802));
j = 1.21551 + (rect * (rect - 2.0580201));
sine = i*j*k*60.252201*x;
full discussion here:
http://synthmaker.co.uk/forum/viewtopic.php?f=4&t=6457&st=0&sk=t&sd=a
I presume that you know, that using a division is a lot slower than multiplying by decimal number, /5 is always slower than *0.2
it's just an approximation.
also:
streamin ramp;
streamin x; // 1.5 = Saw 3.142 = Sin 4.5 = SawSin
streamout sine;
float saw,saw2;
saw = (ramp * 2 - 1) * x;
saw2 = saw * saw;
sine = -0.166667 + saw2 * (0.00833333 + saw2 * (-0.000198409 + saw2 * (2.7526e-006+saw2 * -2.39e-008)));
sine = saw * (1+ saw2 * sine);