In matrix_modn_dense_float.pyx, MAX_MODULUS is set as 2**8 in line 41, and MAX_MODULUS is used to create a Matrix_modn_dense_template object. An overflow error is raised if n >= MAX_MODULUS. However, the docstring states that n values upto 2**11 are supported. The documentation should clearly mention that because LinBox runs really low on values greater than 2**8, it is set as a hard limit.

Interestingly, this is only the case for matrix_modn_dense_float.pyx and not matrix_modn_dense_double.pyx. In the double variant, MAX_MODULUS = 2**23 and is consistent with the docstring and LinBox's upper limit. Is the double variant not slow? Does LinBox only bottleneck on the float variant? I'll do some experiments tomorrow and try to figure this out, but if somebody else has worked with this, do tell.