Anything that uses dbcsr_multiply, i.e. sparse matrix multiplication, when compiled with -D__DBCSR_CUDA. This benefits in particular the linear scaling DFT code. See also the DBCSR project.

If linked against an accelerated scalapack/blas library (in particular pdgemm/pdsyrk/dgemm) that executes these calls on the GPU. The impact of this is most visible for MP2 and RPA calculations. On the hybrid Cray XC30 linking against libsci_acc makes this happen.

FFTs, when compiled with -D__PW_CUDA.