org.apfloat.internal.FloatFactor3NTTStepStrategy Maven / Gradle / Ivy
Show all versions of apfloat Show documentation
package org.apfloat.internal;
import org.apfloat.ApfloatRuntimeException;
import org.apfloat.spi.Factor3NTTStepStrategy;
import org.apfloat.spi.DataStorage;
import static org.apfloat.internal.FloatModConstants.*;
/**
* Steps for the factor-3 NTT.
*
* The transform is done using a parallel algorithm, if the data fits in memory.
*
* All access to this class must be externally synchronized.
*
* @since 1.7.0
* @version 1.8.0
* @author Mikko Tommila
*/
public class FloatFactor3NTTStepStrategy
extends FloatModMath
implements Factor3NTTStepStrategy, Parallelizable
{
// Runnable for transforming the columns in a factor-3 transform
private class ColumnTransformRunnable
implements Runnable
{
public ColumnTransformRunnable(DataStorage dataStorage0, DataStorage dataStorage1, DataStorage dataStorage2, long startColumn, long columns, float w, float ww, float w1, float w2, boolean isInverse)
{
this.dataStorage0 = dataStorage0;
this.dataStorage1 = dataStorage1;
this.dataStorage2 = dataStorage2;
this.startColumn = startColumn;
this.columns = columns;
this.w = w;
this.ww = ww;
this.w1 = w1;
this.w2 = w2;
this.isInverse = isInverse;
}
public void run()
{
float tmp1 = modPow(this.w, (float) this.startColumn),
tmp2 = modPow(this.ww, (float) this.startColumn);
DataStorage.Iterator iterator0 = this.dataStorage0.iterator(DataStorage.READ_WRITE, this.startColumn, this.startColumn + this.columns),
iterator1 = this.dataStorage1.iterator(DataStorage.READ_WRITE, this.startColumn, this.startColumn + this.columns),
iterator2 = this.dataStorage2.iterator(DataStorage.READ_WRITE, this.startColumn, this.startColumn + this.columns);
for (long i = 0; i < this.columns; i++)
{
// 3-point WFTA on the corresponding array elements
float x0 = iterator0.getFloat(),
x1 = iterator1.getFloat(),
x2 = iterator2.getFloat(),
t;
if (this.isInverse)
{
// Multiply before transform
x1 = modMultiply(x1, tmp1);
x2 = modMultiply(x2, tmp2);
}
// Transform columns
t = modAdd(x1, x2);
x2 = modSubtract(x1, x2);
x0 = modAdd(x0, t);
t = modMultiply(t, this.w1);
x2 = modMultiply(x2, this.w2);
t = modAdd(t, x0);
x1 = modAdd(t, x2);
x2 = modSubtract(t, x2);
if (!this.isInverse)
{
// Multiply after transform
x1 = modMultiply(x1, tmp1);
x2 = modMultiply(x2, tmp2);
}
iterator0.setFloat(x0);
iterator1.setFloat(x1);
iterator2.setFloat(x2);
iterator0.next();
iterator1.next();
iterator2.next();
tmp1 = modMultiply(tmp1, this.w);
tmp2 = modMultiply(tmp2, this.ww);
}
}
private DataStorage dataStorage0;
private DataStorage dataStorage1;
private DataStorage dataStorage2;
private long startColumn;
private long columns;
private float w;
private float ww;
private float w1;
private float w2;
private boolean isInverse;
}
/**
* Default constructor.
*/
public FloatFactor3NTTStepStrategy()
{
}
public void transformColumns(DataStorage dataStorage0, DataStorage dataStorage1, DataStorage dataStorage2, long startColumn, long columns, long power2length, long length, boolean isInverse, int modulus)
throws ApfloatRuntimeException
{
// Transform length is three times a power of two
assert (length == 3 * power2length);
ParallelRunnable parallelRunnable = createColumnTransformParallelRunnable(dataStorage0, dataStorage1, dataStorage2, startColumn, columns, power2length, length, isInverse, modulus);
if (columns <= Integer.MAX_VALUE && // Only if the size fits in an integer, but with memory arrays it should
dataStorage0.isCached() && // Only if the data storage supports efficient parallel random access
dataStorage1.isCached() &&
dataStorage2.isCached())
{
ParallelRunner.runParallel(parallelRunnable);
}
else
{
parallelRunnable.run(); // Just run in current thread without parallelization
}
}
public long getMaxTransformLength()
{
return MAX_TRANSFORM_LENGTH;
}
/**
* Create a ParallelRunnable object for transforming the columns of the matrix
* using a 3-point NTT transform.
*
* @param dataStorage0 The data of the first column.
* @param dataStorage1 The data of the second column.
* @param dataStorage2 The data of the third column.
* @param startColumn The starting element index in the data storages to transform.
* @param columns How many columns to transform.
* @param power2length Length of the column transform.
* @param length Length of total transform (three times the length of one column).
* @param isInverse true if an inverse transform is performed, false if a forward transform is performed.
* @param modulus Index of the modulus.
*
* @return A suitable object for performing the 3-point transforms in parallel.
*/
protected ParallelRunnable createColumnTransformParallelRunnable(final DataStorage dataStorage0, final DataStorage dataStorage1, final DataStorage dataStorage2, final long startColumn, final long columns, long power2length, long length, final boolean isInverse, int modulus)
{
setModulus(MODULUS[modulus]); // Modulus
final float w = (isInverse ?
getInverseNthRoot(PRIMITIVE_ROOT[modulus], length) :
getForwardNthRoot(PRIMITIVE_ROOT[modulus], length)), // Forward/inverse n:th root
w3 = modPow(w, (float) power2length), // Forward/inverse 3rd root
ww = modMultiply(w, w),
w1 = negate(modDivide((float) 3, (float) 2)),
w2 = modAdd(w3, modDivide((float) 1, (float) 2));
ParallelRunnable parallelRunnable = new ParallelRunnable(columns)
{
public Runnable getRunnable(long strideStartColumn, long strideColumns)
{
return new ColumnTransformRunnable(dataStorage0, dataStorage1, dataStorage2, startColumn + strideStartColumn, strideColumns, w, ww, w1, w2, isInverse);
}
};
return parallelRunnable;
}
}