Chris@19: Chris@19: Chris@19: Chris@19: FFTW FAQ - Section 5 Chris@19: Chris@19: Chris@19: Chris@19: Chris@19: Chris@19:

Chris@19: FFTW FAQ - Section 5
Chris@19: Known bugs Chris@19:

Chris@19: Chris@19:
Chris@19: Chris@19:

Chris@19: Question 5.1. FFTW 1.1 crashes in rfftwnd on Chris@19: Linux. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 1.2. There was a bug in Chris@19: rfftwnd causing an incorrect amount of memory to be allocated. The bug showed Chris@19: up in Linux with libc-5.3.12 (and nowhere else that we know of). Chris@19: Chris@19:

Chris@19: Question 5.2. The MPI transforms in FFTW 1.2 give incorrect Chris@19: results/leak memory. Chris@19:

Chris@19: Chris@19: These bugs were corrected in FFTW 1.2.1. The MPI transforms (really, Chris@19: just the transpose routines) in FFTW 1.2 had bugs that could cause Chris@19: errors in some situations. Chris@19:

Chris@19: Question 5.3. The test programs in FFTW 1.2.1 fail when I change FFTW Chris@19: to use single precision. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 1.3. (Older versions of FFTW did Chris@19: work in single precision, but the test programs didn't--the error Chris@19: tolerances in the tests were set for double precision.) Chris@19: Chris@19:

Chris@19: Question 5.4. The test program in FFTW 1.2.1 fails for n > Chris@19: 46340. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 1.3. FFTW 1.2.1 produced the right answer, Chris@19: but the test program was wrong. For large n, n*n in the naive Chris@19: transform that we used for comparison overflows 32 bit integer Chris@19: precision, breaking the test. Chris@19:

Chris@19: Question 5.5. The threaded code fails on Linux Redhat Chris@19: 5.0 Chris@19:

Chris@19: Chris@19: We had problems with glibc-2.0.5. The code should work with Chris@19: glibc-2.0.7. Chris@19:

Chris@19: Question 5.6. FFTW 2.0's rfftwnd fails for rank > 1 transforms Chris@19: with a final dimension >= 65536. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 2.0.1. (There was a 32-bit integer Chris@19: overflow due to a poorly-parenthesized expression.) Chris@19:

Chris@19: Question 5.7. FFTW 2.0's complex transforms give the wrong results Chris@19: with prime factors 17 to 97. Chris@19:

Chris@19: Chris@19: There was a bug in the complex transforms that could cause incorrect Chris@19: results under (hopefully rare) circumstances for lengths with Chris@19: intermediate-size prime factors (17-97). This bug was fixed in FFTW Chris@19: 2.1.1. Chris@19:

Chris@19: Question 5.8. FFTW 2.1.1's MPI test programs crash with Chris@19: MPICH. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 2.1.2. The 2.1/2.1.1 MPI test programs Chris@19: crashed when using the MPICH implementation of MPI with the Chris@19: ch_p4 device (TCP/IP); the transforms themselves worked fine. Chris@19: Chris@19:

Chris@19: Question 5.9. FFTW 2.1.2's multi-threaded transforms don't work on Chris@19: AIX. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 2.1.3. The multi-threaded transforms in Chris@19: previous versions didn't work with AIX's Chris@19: pthreads implementation, which idiosyncratically creates threads in detached Chris@19: (non-joinable) mode by default. Chris@19:

Chris@19: Question 5.10. FFTW 2.1.2's complex transforms give incorrect results Chris@19: for large prime sizes. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 2.1.3. FFTW's complex-transform algorithm Chris@19: for prime sizes (in versions 2.0 to 2.1.2) had an integer overflow Chris@19: problem that caused incorrect results for many primes greater than Chris@19: 32768 (on 32-bit machines). (Sizes without large prime factors are Chris@19: not affected.) Chris@19:

Chris@19: Question 5.11. FFTW 2.1.3's multi-threaded transforms don't give any Chris@19: speedup on Solaris. Chris@19:

Chris@19: Chris@19: This bug was fixed in FFTW 2.1.4. (By default, Solaris creates Chris@19: threads that do not parallelize over multiple processors, so one has Chris@19: to request the proper behavior specifically.) Chris@19: Chris@19:

Chris@19: Question 5.12. FFTW 2.1.3 crashes on AIX. Chris@19:

Chris@19: Chris@19: The FFTW 2.1.3 configure script picked incorrect compiler flags for the xlc compiler on newer IBM processors. This Chris@19: is fixed in FFTW 2.1.4.
Chris@19: Back: Internals of FFTW.
Chris@19: Return to contents.

Chris@19:

Chris@19: Matteo Frigo and Steven G. Johnson / fftw@fftw.org Chris@19: - 04 March 2014 Chris@19:

Chris@19: Extracted from FFTW Frequently Asked Questions with Answers, Chris@19: Copyright © 2014 Matteo Frigo and Massachusetts Institute of Technology. Chris@19: