<?xml version="1.0" encoding="utf-8"?>
<feed xml:lang="en" xmlns="http://www.w3.org/2005/Atom"><title>Recent changes to bugs</title><link href="https://sourceforge.net/p/jtransforms/bugs/" rel="alternate"/><link href="https://sourceforge.net/p/jtransforms/bugs/feed.atom" rel="self"/><id>https://sourceforge.net/p/jtransforms/bugs/</id><updated>2012-02-16T08:15:56Z</updated><subtitle>Recent changes to bugs</subtitle><entry><title>Disabled concurrency in FloatFFT_2D for size &gt;= 64K</title><link href="https://sourceforge.net/p/jtransforms/bugs/6/" rel="alternate"/><published>2012-02-16T08:15:56Z</published><updated>2012-02-16T08:15:56Z</updated><author><name>Laurent Bourgès</name><uri>https://sourceforge.net/u/llooo/</uri></author><id>https://sourceforge.netae178a29a41ae5dd6b9400045408c5d79ad07902</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;This is related to the FloatFFT_2D constructor that defines the useThreads variable: rows x columns &amp;gt; java int maximum values (32-bit signed 2,147,483,647)&lt;/p&gt;
&lt;p&gt;Here is the trivial fix (convert int to long):&lt;/p&gt;
&lt;p&gt;/**&lt;br /&gt;
* Creates new instance of FloatFFT_2D.&lt;br /&gt;
*&lt;br /&gt;
* @param rows number of rows&lt;br /&gt;
* @param columns number of columns&lt;br /&gt;
*/&lt;br /&gt;
public FloatFFT_2D(int rows, int columns) {&lt;br /&gt;
if (rows &amp;lt;= 1 || columns &amp;lt;= 1) {&lt;br /&gt;
throw new IllegalArgumentException("rows and columns must be greater than 1");&lt;br /&gt;
}&lt;br /&gt;
this.rows = rows;&lt;br /&gt;
this.columns = columns;&lt;/p&gt;
&lt;p&gt;// fix integer capacity for rows = columns = 65536 !&lt;br /&gt;
if (((long)rows * columns) &amp;gt;= ConcurrencyUtils.getThreadsBeginN_2D()) {&lt;br /&gt;
this.useThreads = true;&lt;br /&gt;
}&lt;br /&gt;
...&lt;br /&gt;
}&lt;/p&gt;
&lt;p&gt;Could you check and fix that bug anywhere it could happen (xxxFFT_2D, xxxFFT_3D ...) ?&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>FloatFFT_2D.realForward(float[][]) hangs for size &gt;= 8K </title><link href="https://sourceforge.net/p/jtransforms/bugs/5/" rel="alternate"/><published>2012-02-16T08:14:50Z</published><updated>2012-02-16T08:14:50Z</updated><author><name>Laurent Bourgès</name><uri>https://sourceforge.net/u/llooo/</uri></author><id>https://sourceforge.net5031a19090ade990417faed5359778fee7cd5e39</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;This is related to FFT 1D concurrency thresholds whereas all threads are already busy within FloatFFT_2D.realForward().&lt;/p&gt;
&lt;p&gt;As a workaround, I explicitly set these thresholds to INFINITY to disable FFT 1D parallelism:&lt;br /&gt;
// JTransforms: disable FFT_1D parallelism:&lt;br /&gt;
ConcurrencyUtils.setThreadsBeginN_1D_FFT_2Threads(Integer.MAX_VALUE);&lt;br /&gt;
ConcurrencyUtils.setThreadsBeginN_1D_FFT_4Threads(Integer.MAX_VALUE);&lt;/p&gt;
&lt;p&gt;Maybe JTransforms could automatically disable FFT 1D parallelism when FFT 2D are computed ... &lt;/p&gt;
&lt;p&gt;I propose that the FloatFFT_2D constructor which instanciate FloatFFT_1D objects (row/columns) define a flag useThreads = false in order to disable concurrency in such cases ...&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>[rev 25] FloatFFT_2D.realForwardFull of data with 6 columns</title><link href="https://sourceforge.net/p/jtransforms/bugs/4/" rel="alternate"/><published>2011-07-06T02:49:21Z</published><updated>2011-07-06T02:49:21Z</updated><author><name/><uri>https://sourceforge.net</uri></author><id>https://sourceforge.neta1af58ace5803b785a17c3834c9f284228abbf51</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;This is the same bug as #3355237, for the single precision (float) version of the 2d-FFT.&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>[rev 25] DoubleFFT_2D.realForwardFull of data with 6 columns</title><link href="https://sourceforge.net/p/jtransforms/bugs/3/" rel="alternate"/><published>2011-07-06T02:47:30Z</published><updated>2011-07-06T02:47:30Z</updated><author><name/><uri>https://sourceforge.net</uri></author><id>https://sourceforge.netcf447e658f3a64952ae066d4c04ee99c46648f65</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;The following unit tests fail (see attached report)&lt;br /&gt;
- DoubleFFT_2DTest.testRealForwardFull1dinput&lt;br /&gt;
- DoubleFFT_2DTest.testRealInverseFullScaled1dinput&lt;br /&gt;
- DoubleFFT_2DTest.testRealInverseFullUnscaled1dinput&lt;br /&gt;
for the following data&lt;br /&gt;
- number of rows = any&lt;br /&gt;
- number of columns = 6&lt;br /&gt;
- multithreaded version of the algorithm.&lt;br /&gt;
Note that the first wrong value corresponds to index [6 * rows + 1]. Expected value 0. It seems that data with number of rows which are a multiple of 6 are not affected.&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>[rev 23] FloatFFT_3D.realForward fails the unit tests</title><link href="https://sourceforge.net/p/jtransforms/bugs/2/" rel="alternate"/><published>2011-07-05T07:12:15Z</published><updated>2011-07-05T07:12:15Z</updated><author><name/><uri>https://sourceforge.net</uri></author><id>https://sourceforge.net533ba07887534d669d0416cb514953ff283f8483</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;All invocations of FloatFFT_3DTest.testRealForward1fInput and FloatFFT_3DTest.testRealForward3fInput fail (see attached report).&lt;/p&gt;&lt;/div&gt;</summary></entry><entry><title>3d real FFT fails</title><link href="https://sourceforge.net/p/jtransforms/bugs/1/" rel="alternate"/><published>2011-07-02T16:35:37Z</published><updated>2011-07-02T16:35:37Z</updated><author><name/><uri>https://sourceforge.net</uri></author><id>https://sourceforge.nete57aad6ffe02d96fc19476b036350135d1cb36c2</id><summary type="html">&lt;div class="markdown_content"&gt;&lt;p&gt;Unit tests&lt;br /&gt;
- DoubleFFT_3DTest.testRealForward(double[]),&lt;br /&gt;
- DoubleFFT_3DTest.testRealForward(double[][][]),&lt;br /&gt;
both fail. This a a real bug, not an accuracy issue.&lt;/p&gt;&lt;/div&gt;</summary></entry></feed>