com.hazelcast.util

Class HashUtil

java.lang.Object

com.hazelcast.util.HashUtil

public final class HashUtil
extends Object

Utility methods related to hashtables.

Method Summary

Methods

Modifier and Type	Method and Description
static int	computePerturbationValue(int capacity) Compute the key perturbation value applied before hashing.
static int	fastIntMix(int k)
static long	fastLongMix(long k)
static int	hashCode(Object... objects) Hash code for multiple objects using Arrays.hashCode(Object[]).
static int	hashToIndex(int hash, int length) A function that calculates the index (e.g.
static int	MurmurHash3_fmix(int k)
static long	MurmurHash3_fmix(long k)
static long	MurmurHash3_x64_64_direct(long address, int offset, int len)
static long	MurmurHash3_x64_64_direct(long address, int offset, int len, int seed)
static long	MurmurHash3_x64_64(byte[] data, int offset, int len)
static long	MurmurHash3_x64_64(byte[] data, int offset, int len, int seed)
static int	MurmurHash3_x86_32_direct(long address, int offset, int len)
static int	MurmurHash3_x86_32_direct(long address, int offset, int len, int seed) Returns the MurmurHash3_x86_32 hash.
static int	MurmurHash3_x86_32(byte[] data, int offset, int len)
static int	MurmurHash3_x86_32(byte[] data, int offset, int len, int seed) Returns the MurmurHash3_x86_32 hash.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

MurmurHash3_x86_32

public static int MurmurHash3_x86_32(byte[] data,
                     int offset,
                     int len)

MurmurHash3_x86_32

public static int MurmurHash3_x86_32(byte[] data,
                     int offset,
                     int len,
                     int seed)

Returns the MurmurHash3_x86_32 hash.

MurmurHash3_x86_32_direct

public static int MurmurHash3_x86_32_direct(long address,
                            int offset,
                            int len)

MurmurHash3_x86_32_direct

public static int MurmurHash3_x86_32_direct(long address,
                            int offset,
                            int len,
                            int seed)

Returns the MurmurHash3_x86_32 hash.

MurmurHash3_x64_64

public static long MurmurHash3_x64_64(byte[] data,
                      int offset,
                      int len)

MurmurHash3_x64_64

public static long MurmurHash3_x64_64(byte[] data,
                      int offset,
                      int len,
                      int seed)

MurmurHash3_x64_64_direct

public static long MurmurHash3_x64_64_direct(long address,
                             int offset,
                             int len)

MurmurHash3_x64_64_direct

public static long MurmurHash3_x64_64_direct(long address,
                             int offset,
                             int len,
                             int seed)

MurmurHash3_fmix

public static int MurmurHash3_fmix(int k)

MurmurHash3_fmix

public static long MurmurHash3_fmix(long k)

fastLongMix

public static long fastLongMix(long k)

fastIntMix

public static int fastIntMix(int k)

hashCode

public static int hashCode(Object... objects)

Hash code for multiple objects using Arrays.hashCode(Object[]).

hashToIndex

public static int hashToIndex(int hash,
              int length)

A function that calculates the index (e.g. to be used in an array/list) for a given hash. The returned value will always be equal or larger than 0 and will always be smaller than 'length'. The reason this function exists is to deal correctly with negative and especially the Integer.MIN_VALUE; since that can't be used safely with a Math.abs function.

Parameters:

length - the length of the array/list

Returns:

the mod of the hash

Throws:

IllegalArgumentException - if mod smaller than 1.

computePerturbationValue

public static int computePerturbationValue(int capacity)

Compute the key perturbation value applied before hashing. The returned value should be non-zero and ideally different for each capacity. This matters because keys are nearly-ordered by their hashed values so when adding one container's values to the other, the number of collisions can skyrocket into the worst case possible.

If it is known that hash containers will not be added to each other (will be used for counting only, for example) then some speed can be gained by not perturbing keys before hashing and returning a value of zero for all possible capacities. The speed gain is a result of faster rehash operation (keys are mostly in order).