FAME ON GPU

For a better computational efficiency, we implemented the FAME on the GPU device. From the CUDA-based programs, users can get more than a hundred times the efficiency of the CPU. By the way, user only needs to set the appropriate parameters and run the executable file can quickly get the results.

Download executable files

Cubic system

  • Simple cubic (version 3.0)
  • Face centered cubic (version 3.5)
  • Body centered cubic (version 3.5)

Environmental requirements

  • Cuda (version 8.0)
  • Lapack (version 3.6.0)

How to start FAME on GPU device?

1. Download the executable file and parameter setting files.

2. Place these files under a subdirectory and perform the following action (e.g. FCC case) :

$ ./PC_FCC.out PC_para_FCC.txt LSEV_info.txt

3. If it is working properly, you will see what as following figure.

4. When this program is finished, all the eigenvalues of the result of this computation will be in a file named 'eigenval.txt'.

How to use FAME on GPU device?

The current FAME implementation on the GPU is divided into the following versions depending on the lattice type. Notice that only isotropic material type could be choose now (an-isotropic and bi-isotropic are invalid now on GPU device).