| 9:52:35 AM PDT - Tue, Mar 19th 2019 |
|
Quote:Ftorazyne Mar 18th 11:39 am
Thank you for you reply and advices. As I understand, BLAS_SIZE=4 means NwChem compiled with single precision floats, so I should to compile OpenBLAS with single precision float too (I compiled with double precision). Is there significant difference between single precision and double precision NwChem?
No, that does not refer to floating-point precision, but to the size of integer arguments in BLAS calls.
BLAS_SIZE=4 stands for 32-bit integers (most common default )
BLAS_SIZE=8 stands for 64-bit integers
|