blfoat(gamma(3/4)/gamma(1/4)) returns 4.125843750410604b-5. The correct answer should be close to float(gamma(3/4)/gamma(1/4) = .3379891200336423.
The issue is caused by the fact that bffac uses 4 extra digits in the computation and that result is returned. It eventually gets used in invertbigfloat which divides 1 by this value. The precisions don't match which causes the problem.
Log in to post a comment.