[SciPy-User] scipy 64bit cannot handle very large matrix (integer overflow)

[Sturla Molden]

Den 30. jan. 2013 kl. 01:44 skrev osman buyukisik <osman at fuse.net>:

> 
> Isn't this usually some kind of compile option like "-i8"  ??
> 

Sometimes — i.e. integers without kind number in the code. Sometimes the kind number is hard-coded. Sometimes the kind number is computed by calling the function "selected_int_kind".

On x64, Fortran libraries are usually compled to use 32-bit integers, as it is the native integer size  on the platform (64 bit address with 32 bit offset). That is also the reason why a C long on Windows 64 is 32 bits.

Sturla