k-Wave

1. Karthik Nagabhushana
   

   Member
   Joined: Apr '20
   Posts: 6

   Hi,

   Is it possible to set the GPU binaries based simulation to be in 'double' precision? I currently generate the input HDF file using 'kspaceFirstOrder3D(kgrid, medium, source, sensor, 'SaveToDisk', filename, input_args{:})'. I tried setting 'DataCast' option to 'double' and 'gpuArray-double'. But the the output variables were always in single precision.

   Thanks,
   Karthik

   Posted 2 years ago #

2. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   No - the C++ binaries work in single precision only (the data cast options are not used when saving to disk). Actually, there are very few cases when computing in double precision is needed. The accuracy of most simulations will be limited by the accuracy of the PML, which is a few decimal points.

   Posted 2 years ago #

3. Karthik Nagabhushana
   

   Member
   Joined: Apr '20
   Posts: 6

   Got it. Thank you, Professor Treeby.

   Posted 2 years ago #

RSS feed for this topic

Reply

You must log in to post.