Force single Precision Calculation for Cuda
Posted: Thu Jan 05, 2017 8:51 am
Hi, I would like to use the single precision calculation for GPU in Abinit, however, I get an error running with
This is the error Message
Can somebody help me force single precision?
Code: Select all
use_gpu_cudaThis is the error Message
Code: Select all
--- !ERROR
src_file: invars0.F90
src_line: 431
mpi_rank: 0
message: |
Input variables use_gpu_cuda is on but abinit hasn't been built
with (double precision) gpu mode enabled !
Action : change the input variable use_gpu_cuda
or re-compile ABINIT with double-precision Cuda enabled.Can somebody help me force single precision?