In message <53c230b615chr...@care4free.net> Chris Gransden <chr...@care4free.net> wrote:
> In article <7d8929c253.b...@ron1954.woosh.co.nz>, > Ron <b...@woosh.co.nz> wrote: > > It is the latest libsndfile test-suite that I am running on my Iyonix. > > I didn't use the autobuilder but just the cross-compiler tools. > > There is provision to produce a cross-compiler test-tarball which almost > > works except for an elf extension bungle on one line. > > With a bit of tinkering I got it to go, and made the script into an obey > > file. > > The test for pipe still expects a unix cat command and pipe, and there > > is several file open errors, but the scale-float test is showing a poor > > result, almost too poor, as the libsndfile based resampler prog appears > > to work OK. Here is a link to the text output > > http://homepages.woosh.co.nz/ron.may/sndfile-test.txt > > I have been using the latest source by git, the earlier tarballs were'nt > > opening the wav files at all. > > I ran the one in the autobuilder, which is 1.0.25. The tests for > 'floating_point_test' all passed ok. I used both versions of the compiler > in the autobuilder, gcc 4.1.2 with softfloat and gcc 4.7.4 with vfp. > The vfp version ran 15x faster than softfloat on a Pandaboard. > > Chris. > Wow, that's encouraging, one google revealed the author of libsndfle claiming some processors like arm did not have lrintf so I was imagining an incompatibility. That is the same floating_point_test by the looks. The other thing I was thinking was if it was scaled against time it would give a poor result, but if you have had it passing that's good. In the autobuilder I have spotted '--disable-external-libs' so I'll try that before going to the autobuilder libsndfile in a last effort. It is good to hear that gcc 4.7.4 and vfp is working so well, is the Pandaboard the only one that can benefit? Ron M. _______________________________________________ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK