Hm. fftw A34 is still not equal to FFTA34. On win32 I see the same failure with the old script, though (and the new script gives the same answer for fftw A34 as the old script did).
Have you tried this yourself? Does the test pass for you (with either old or new script)? David "bill lam" <[email protected]> wrote in message news:[email protected]... > dim, 21 Mar 2010, David Hotham skribis: >> The tests below pass with the following exceptions: >> >> - those involving A34. I fancy we're getting different random values. >> Certainly I get different random values on w32 than I do on w64 (also >> they're different machines; I don't know which difference is important). >> Therefore FFTA34 doesn't match. > > could you test with A34 fixed > > A34=: ". ;._2 (0 : 0) > 0j2 1j6 2j1 0j12 > 4j11 3j6 0j5 1j2 > 4j5 4j14 5j1 2j12 > ) > > >> - The interface has changed enough in FFTW3 that the final set of tests >> are >> not correct >> - the in/out parameters are now passed to createplan, and fftwndone is >> replaced by fftwexecute > > Could you modify the section to use the new api and test again. I've > never used fftw but I hope fftw3 can pass test script before commit to > svn. > > -- > regards, > ==================================================== > GPG key 1024D/4434BAB3 2008-08-24 > gpg --keyserver subkeys.pgp.net --recv-keys 4434BAB3 > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
