Fastest implementation of log2(int) and log2(float)

Question

The question is

Are there any other (and/or faster) implementations of a basic 2log?

Applications

The log2(int) and

Answer 1

There have been quite a few answers providing fast approximate approaches to log2(int) but few for log2(float), so here's two (Java implementation given) that use both a lookup table and mantissa/bit hacking:

Fast accurate log2(float):

/**
 * Calculate the logarithm to base 2, handling special cases.
 */
public static float log2(float x) {

    final int bits = Float.floatToRawIntBits(x);
    final int e = (bits >> 23) & 0xff;
    final int m = (bits & 0x7fffff);

    if (e == 255) {
        if (m != 0) {
            return Float.NaN;
        }
        return ((bits >> 31) != 0) ? Float.NaN : Float.POSITIVE_INFINITY;
    }

    if ((bits >> 31) != 0) {
        return (e == 0 && m == 0) ? Float.NEGATIVE_INFINITY : Float.NaN;
    }

    return (e == 0 ? data[m >>> qm1] : e + data[((m | 0x00800000) >>> q)]);
}

Note:

• If the argument is NaN or less than zero, then the result is NaN.
• If the argument is positive infinity, then the result is positive infinity.
• If the argument is positive zero or negative zero, then the result is negative infinity.

Fast accurate log2(float) (slightly faster, no checking):

/**
 * Calculate the logarithm using base 2. Requires the argument be finite and
 * positive.
 */
public static float fastLog2(float x) {
    final int bits = Float.floatToRawIntBits(x);
    final int e = (bits >> 23) & 0xff;
    final int m = (bits & 0x7fffff);
    return (e == 0 ? data[m >>> qm1] : e + data[((m | 0x00800000) >>> q)]);
}

This second method forgoes the checking present in the other method and therefore has the following special cases:

• If the argument is NaN, then the result is incorrect.
• If the argument is negative, then the result is incorrect.
• If the argument is positive infinity, then the result is incorrect.
• If the argument is positive zero or negative zero, then the result is negative infinity.

---

Both methods upon rely on a lookup table data (and variables q and qm1). These are populated with the following method. n defines the accuracy-space tradeoff.

static int q, qm1;
static float[] data;

/**
 * Compute lookup table for a given base table size.
 * 
 * @param n The number of bits to keep from the mantissa. Table storage =
 *          2^(n+1) * 4 bytes, e.g. 64Kb for n=13. Must be in the range
 *          0<=n<=23
 */
public static void populateLUT(int n) {

    final int size = 1 << (n + 1);

    q = 23 - n;
    qm1 = q - 1;
    data = new float[size];

    for (int i = 0; i < size; i++) {
        data[i] = (float) (Math.log(i << q) / Math.log(2)) - 150;
    }
}

---

populateLUT(12);
log2(6666); // = 12.702606

Answer 2

(I haven't done any measurements so this may not match up, but I thought user user9337139's idea was neat and wanted to try the same in C# - his is C++).

Here's a C# int Magnitude(byte) function based on converting the byte value to float and extracting the exponent from the IEEE float representation.

    using System.Runtime.InteropServices;

    [StructLayout(LayoutKind.Explicit)]
    struct UnionWorker
    {
        [FieldOffset(0)]
        public int i;
        [FieldOffset(0)]
        public float f;
    }

    static int Magnitude(byte b)
    {
        UnionWorker u;
        u.i = 0; // just to please the compiler
        u.f = b;
        return Math.Max((u.i >> 23) & 0xFF, 126) - 126;
    }

Returns zero for zero, 8 for 0xFF, other values as you would expect.

Zero is a special case, so I needed the Math.Max clamp for that. I suspect user9337139's solution might have a similar problem.

Note, this has not been tested for endianness issues - caveat emptor.

Answer 3

For more algorithms look here http://www.asmcommunity.net/forums/topic/?id=15010

Also did some testing in C++ and my implementation of BSR is slower than lookup table

• i am using BDS2006 there is probably slow down by state push/popping by asm directive
• your lookup is fine but i am using 11 bits table instead of 8
• it divides 32 bit into 3 branches instead of 4
• and it is still small enough to handle without init function

code:

//---------------------------------------------------------------------------
DWORD log2_slow(const DWORD &x)
    {
    DWORD m,i;
    if (!x) return 0;
    if (x>=0x80000000) return 31;
    for (m=1,i=0;m<x;m<<=1,i++);
     if (m!=x) i--;
    return i;
    }
//---------------------------------------------------------------------------
DWORD log2_asm(const DWORD &x)
    {
    DWORD xx=x;
    asm {
        mov eax,xx
        bsr eax,eax;
        mov xx,eax;
        }
    return xx;
    }
//---------------------------------------------------------------------------
BYTE _log2[2048]=
    {
     0, 0, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
     7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
     8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
     8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
     9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
     9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
     9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
     9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,
    };
DWORD log2(const DWORD &x)
    {
         if (x>=0x00400000) return _log2[x>>22]+22;
    else if (x>=0x00000800) return _log2[x>>11]+11;
    else                    return _log2[x];
    }
//---------------------------------------------------------------------------

test code:

DWORD x,j,i,n=256;
tbeg(); for (i=0;i<32;i++) for (j=0;j<n;j++) x=log2     (j<<i); tend(); mm_log->Lines->Add(tstr(1));
tbeg(); for (i=0;i<32;i++) for (j=0;j<n;j++) x=log2_asm (j<<i); tend(); mm_log->Lines->Add(tstr(1));
tbeg(); for (i=0;i<32;i++) for (j=0;j<n;j++) x=log2_slow(j<<i); tend(); mm_log->Lines->Add(tstr(1));

my results on AMD A8-5500 3.2 GHz:

[   0.040 ms] log2     (x) - 11bit lookup table
[   0.060 ms] log2_asm (x) - BSR
[   0.415 ms] log2_slow(x) - shift loop

Note:

• log2(0) -> 0 because of use of DWORDS, in real it should be -inf
• all other values are correct for all functions

Answer 4

inline int fast_log2(register double x)
{ 
    return (reinterpret_cast<uint64_t&>(x) >> 52) - 1023;
};

Answer 5

There are some integer algorithms here.

In C#:

public static uint FloorLog2(uint x)
{
    x |= (x >> 1);
    x |= (x >> 2);
    x |= (x >> 4);
    x |= (x >> 8);
    x |= (x >> 16);

    return (uint)(NumBitsSet(x) - 1);
}

public static uint CeilingLog2(uint x)
{
    int y = (int)(x & (x - 1));

    y |= -y;
    y >>= (WORDBITS - 1);
    x |= (x >> 1);
    x |= (x >> 2);
    x |= (x >> 4);
    x |= (x >> 8);
    x |= (x >> 16);

    return (uint)(NumBitsSet(x) - 1 - y);
}

public static int NumBitsSet(uint x)
{
    x -= ((x >> 1) & 0x55555555);
    x = (((x >> 2) & 0x33333333) + (x & 0x33333333));
    x = (((x >> 4) + x) & 0x0f0f0f0f);
    x += (x >> 8);
    x += (x >> 16);

    return (int)(x & 0x0000003f);
}

private const int WORDBITS = 32;

You should look at the original code on the site I linked for the context, particularly what happens with Log2(0).

Answer 6

    static byte FloorLog2(UInt16 value)
    {
        for (byte i = 0; i < 15; ++i)
        {
            if ((value >>= 1) < 1)
            {
                return i;
            }
        }
        return 15;
    }