GPU: Add GPU Baseline for ABACUS !

[denghuilu]

CUDA GPU implementations

In ABACUS, we provide the option to use the GPU devices to accelerate the performance. In the tested si16_pw system, we have found an order of magnitude speedup compared with the CPU code.
And it has the following general features:

• Full gpu implementations: During the SCF progress, Psi, Hamilt, Hsolver, and DiagCG classes are stored or calculated by the GPU devices.

• Electronic state data: (e.g. electronic density) are moved from the GPU to the CPU(s) every SCF step.

• Acclerated by the NVIDIA libraries: cuBLAS for common linear algebra calculations, cuSolver for eigen values/vectors, and cuFFT for the conversions between the real and recip spaces.

• Multi GPU supprted: Using multiple MPI tasks will often give the best performance. Note each MPI task will be bind to a GPU device with automatically computing load balancing.

• Parallel strategy: K point parallel.

Building ABACUS with the GPU support:

Check the Advanced Installation Options for the installation of CUDA version support.

Run with the GPU support by editing the INPUT script:

In INPUT file we need to set the value keyword device to be gpu.

Examples

We provides examples of gpu calculations.

Known limitations

• Only CG method is supported, so the input keyword ks_solver can only take the value cg,
• Only PW basis is supported, so the input keyword basis_type can only take the value pw,
• Only k point parallelization is supported, so the input keyword kpar will be set to match the number of MPI tasks automatically.
• Supported CUDA architectures:
  • 60 # P100, 1080ti
  • 70 # V100
  • 75 # T4
  • 80 # A100, 3090