Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/doc/FAQ/fftw-faq.html/section5.html @ 127:7867fa7e1b6b

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