It seems to only be significantly different with fannkuch-redux. Looking at the CLBG page (http://benchmarksgame.alioth.debian.org/u32/performance.php?test=fann

It seems to only be significantly different with fannkuch-redux. Looking at the CLBG page (http://benchmarksgame.alioth.debian.org/u32/performance.php?test=fannkuchredux&sort=elapsed http://benchmarksgame.alioth.debian.org/u32/performance.php?...) for that algorithm there doesn't seem to be very big difference at all. I cannot explain the difference and if this is indeed an error by the authors of the paper it casts a fairly big shadow on the reliability of the rest of the data.