annotate fft/fftw/fftw-3.3.4/doc/FAQ/fftw-faq.html/section5.html @ 40:223f770b5341 kissfft-double tip

Try a double-precision kissfft
author Chris Cannam
date Wed, 07 Sep 2016 10:40:32 +0100
parents 26056e866c29
children
rev   line source
Chris@19 1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
Chris@19 2 <html>
Chris@19 3 <head><title>
Chris@19 4 FFTW FAQ - Section 5
Chris@19 5 </title>
Chris@19 6 <link rev="made" href="mailto:fftw@fftw.org">
Chris@19 7 <link rel="Contents" href="index.html">
Chris@19 8 <link rel="Start" href="index.html">
Chris@19 9 <link rel="Previous" href="section4.html"><link rel="Bookmark" title="FFTW FAQ" href="index.html">
Chris@19 10 </head><body text="#000000" bgcolor="#FFFFFF"><h1>
Chris@19 11 FFTW FAQ - Section 5 <br>
Chris@19 12 Known bugs
Chris@19 13 </h1>
Chris@19 14
Chris@19 15 <ul>
Chris@19 16 <li><a href="#rfftwndbug" rel=subdocument>Q5.1. FFTW 1.1 crashes in rfftwnd on Linux.</a>
Chris@19 17 <li><a href="#fftwmpibug" rel=subdocument>Q5.2. The MPI transforms in FFTW 1.2 give incorrect results/leak
Chris@19 18 memory.</a>
Chris@19 19 <li><a href="#testsingbug" rel=subdocument>Q5.3. The test programs in FFTW 1.2.1 fail when I change FFTW to use single
Chris@19 20 precision.</a>
Chris@19 21 <li><a href="#teststoobig" rel=subdocument>Q5.4. The test program in FFTW 1.2.1 fails for n &gt;
Chris@19 22 46340.</a>
Chris@19 23 <li><a href="#linuxthreads" rel=subdocument>Q5.5. The threaded code fails on Linux Redhat 5.0</a>
Chris@19 24 <li><a href="#bigrfftwnd" rel=subdocument>Q5.6. FFTW 2.0's rfftwnd fails for rank &gt; 1 transforms with a final
Chris@19 25 dimension &gt;= 65536.</a>
Chris@19 26 <li><a href="#primebug" rel=subdocument>Q5.7. FFTW 2.0's complex transforms give the wrong results with prime
Chris@19 27 factors 17 to 97.</a>
Chris@19 28 <li><a href="#mpichbug" rel=subdocument>Q5.8. FFTW 2.1.1's MPI test programs crash with
Chris@19 29 MPICH.</a>
Chris@19 30 <li><a href="#aixthreadbug" rel=subdocument>Q5.9. FFTW 2.1.2's multi-threaded transforms don't work on
Chris@19 31 AIX.</a>
Chris@19 32 <li><a href="#bigprimebug" rel=subdocument>Q5.10. FFTW 2.1.2's complex transforms give incorrect results for large prime
Chris@19 33 sizes.</a>
Chris@19 34 <li><a href="#solaristhreadbug" rel=subdocument>Q5.11. FFTW 2.1.3's multi-threaded transforms don't give any speedup on
Chris@19 35 Solaris.</a>
Chris@19 36 <li><a href="#aixflags" rel=subdocument>Q5.12. FFTW 2.1.3 crashes on AIX.</a>
Chris@19 37 </ul><hr>
Chris@19 38
Chris@19 39 <h2><A name="rfftwndbug">
Chris@19 40 Question 5.1. FFTW 1.1 crashes in rfftwnd on
Chris@19 41 Linux.
Chris@19 42 </A></h2>
Chris@19 43
Chris@19 44 This bug was fixed in FFTW 1.2. There was a bug in
Chris@19 45 <code>rfftwnd</code> causing an incorrect amount of memory to be allocated. The bug showed
Chris@19 46 up in Linux with libc-5.3.12 (and nowhere else that we know of).
Chris@19 47
Chris@19 48 <h2><A name="fftwmpibug">
Chris@19 49 Question 5.2. The MPI transforms in FFTW 1.2 give incorrect
Chris@19 50 results/leak memory.
Chris@19 51 </A></h2>
Chris@19 52
Chris@19 53 These bugs were corrected in FFTW 1.2.1. The MPI transforms (really,
Chris@19 54 just the transpose routines) in FFTW 1.2 had bugs that could cause
Chris@19 55 errors in some situations.
Chris@19 56 <h2><A name="testsingbug">
Chris@19 57 Question 5.3. The test programs in FFTW 1.2.1 fail when I change FFTW
Chris@19 58 to use single precision.
Chris@19 59 </A></h2>
Chris@19 60
Chris@19 61 This bug was fixed in FFTW 1.3. (Older versions of FFTW did
Chris@19 62 work in single precision, but the test programs didn't--the error
Chris@19 63 tolerances in the tests were set for double precision.)
Chris@19 64
Chris@19 65 <h2><A name="teststoobig">
Chris@19 66 Question 5.4. The test program in FFTW 1.2.1 fails for n &gt;
Chris@19 67 46340.
Chris@19 68 </A></h2>
Chris@19 69
Chris@19 70 This bug was fixed in FFTW 1.3. FFTW 1.2.1 produced the right answer,
Chris@19 71 but the test program was wrong. For large n, n*n in the naive
Chris@19 72 transform that we used for comparison overflows 32 bit integer
Chris@19 73 precision, breaking the test.
Chris@19 74 <h2><A name="linuxthreads">
Chris@19 75 Question 5.5. The threaded code fails on Linux Redhat
Chris@19 76 5.0
Chris@19 77 </A></h2>
Chris@19 78
Chris@19 79 We had problems with glibc-2.0.5. The code should work with
Chris@19 80 glibc-2.0.7.
Chris@19 81 <h2><A name="bigrfftwnd">
Chris@19 82 Question 5.6. FFTW 2.0's rfftwnd fails for rank &gt; 1 transforms
Chris@19 83 with a final dimension &gt;= 65536.
Chris@19 84 </A></h2>
Chris@19 85
Chris@19 86 This bug was fixed in FFTW 2.0.1. (There was a 32-bit integer
Chris@19 87 overflow due to a poorly-parenthesized expression.)
Chris@19 88 <h2><A name="primebug">
Chris@19 89 Question 5.7. FFTW 2.0's complex transforms give the wrong results
Chris@19 90 with prime factors 17 to 97.
Chris@19 91 </A></h2>
Chris@19 92
Chris@19 93 There was a bug in the complex transforms that could cause incorrect
Chris@19 94 results under (hopefully rare) circumstances for lengths with
Chris@19 95 intermediate-size prime factors (17-97). This bug was fixed in FFTW
Chris@19 96 2.1.1.
Chris@19 97 <h2><A name="mpichbug">
Chris@19 98 Question 5.8. FFTW 2.1.1's MPI test programs crash with
Chris@19 99 MPICH.
Chris@19 100 </A></h2>
Chris@19 101
Chris@19 102 This bug was fixed in FFTW 2.1.2. The 2.1/2.1.1 MPI test programs
Chris@19 103 crashed when using the MPICH implementation of MPI with the
Chris@19 104 <code>ch_p4</code> device (TCP/IP); the transforms themselves worked fine.
Chris@19 105
Chris@19 106 <h2><A name="aixthreadbug">
Chris@19 107 Question 5.9. FFTW 2.1.2's multi-threaded transforms don't work on
Chris@19 108 AIX.
Chris@19 109 </A></h2>
Chris@19 110
Chris@19 111 This bug was fixed in FFTW 2.1.3. The multi-threaded transforms in
Chris@19 112 previous versions didn't work with AIX's
Chris@19 113 <code>pthreads</code> implementation, which idiosyncratically creates threads in detached
Chris@19 114 (non-joinable) mode by default.
Chris@19 115 <h2><A name="bigprimebug">
Chris@19 116 Question 5.10. FFTW 2.1.2's complex transforms give incorrect results
Chris@19 117 for large prime sizes.
Chris@19 118 </A></h2>
Chris@19 119
Chris@19 120 This bug was fixed in FFTW 2.1.3. FFTW's complex-transform algorithm
Chris@19 121 for prime sizes (in versions 2.0 to 2.1.2) had an integer overflow
Chris@19 122 problem that caused incorrect results for many primes greater than
Chris@19 123 32768 (on 32-bit machines). (Sizes without large prime factors are
Chris@19 124 not affected.)
Chris@19 125 <h2><A name="solaristhreadbug">
Chris@19 126 Question 5.11. FFTW 2.1.3's multi-threaded transforms don't give any
Chris@19 127 speedup on Solaris.
Chris@19 128 </A></h2>
Chris@19 129
Chris@19 130 This bug was fixed in FFTW 2.1.4. (By default, Solaris creates
Chris@19 131 threads that do not parallelize over multiple processors, so one has
Chris@19 132 to request the proper behavior specifically.)
Chris@19 133
Chris@19 134 <h2><A name="aixflags">
Chris@19 135 Question 5.12. FFTW 2.1.3 crashes on AIX.
Chris@19 136 </A></h2>
Chris@19 137
Chris@19 138 The FFTW 2.1.3 <code>configure</code> script picked incorrect compiler flags for the <code>xlc</code> compiler on newer IBM processors. This
Chris@19 139 is fixed in FFTW 2.1.4. <hr>
Chris@19 140 Back: <a href="section4.html" rev=precedes>Internals of FFTW</a>.<br>
Chris@19 141 <a href="index.html" rev=subdocument>Return to contents</a>.<p>
Chris@19 142 <address>
Chris@19 143 <A href="http://www.fftw.org">Matteo Frigo and Steven G. Johnson</A> / <A href="mailto:fftw@fftw.org">fftw@fftw.org</A>
Chris@19 144 - 04 March 2014
Chris@19 145 </address><br>
Chris@19 146 Extracted from FFTW Frequently Asked Questions with Answers,
Chris@19 147 Copyright &copy; 2014 Matteo Frigo and Massachusetts Institute of Technology.
Chris@19 148 </body></html>