README

Multi-GPU molecular dynamics engine for NVIDIA GPU cards.
(c) Antti-Pekka Hynninen, 2013-2014

Requirements:
--------------
* GNU/Intel C++ compiler
* MPI library for multi-GPU support
* CUDA 5.0 or newer
* Compute capability 2.0 (Tesla) or higher NVIDIA GPU
* When compiled with USE_TEXTURE_OBJECTS, compute capability 3.0 (Kepler) or higher required

Compiling:
-----------
make

Running tests:
---------------

gpu_direct
- Test direct space non-bonded force calculation

gpu_bonded
- Test bonded force calculation

gpu_recip
- Test reciprocal space non-bonded force calculation

gpu_const
- Test holonomic constraints (i.e. "SHAKE")

gpu_dyna -nstep N
- Test molecular dynamics run
- Uses MPI for multi-GPU support

Unused code:
-------------
These are classes that are half-written and not yet integrated into the engine
-CudaMonteCarloBaroStat
-LangevinPiston