GitHub - ap-hynninen/gpu_nonbond: Molecular Dynamics on NVIDIA GPUs

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 200 Commits
fbfft		fbfft
temp		temp
test_data		test_data
.gitignore		.gitignore
AtomGroupBase.h		AtomGroupBase.h
Bonded_struct.h		Bonded_struct.h
Bspline.cu		Bspline.cu
Bspline.h		Bspline.h
CpuMatrix3d.cpp		CpuMatrix3d.cpp
CpuMatrix3d.h		CpuMatrix3d.h
CpuMultiNodeMatrix3d.cpp		CpuMultiNodeMatrix3d.cpp
CpuMultiNodeMatrix3d.h		CpuMultiNodeMatrix3d.h
CudaAtomGroup.h		CudaAtomGroup.h
CudaBlock.cu		CudaBlock.cu
CudaBlock.h		CudaBlock.h
CudaBondedForce.cu		CudaBondedForce.cu
CudaBondedForce.h		CudaBondedForce.h
CudaDirectForceKernels.cu		CudaDirectForceKernels.cu
CudaDirectForceKernels.h		CudaDirectForceKernels.h
CudaDirectForceTypes.h		CudaDirectForceTypes.h
CudaDirectForce_util.h		CudaDirectForce_util.h
CudaDomdec.cu		CudaDomdec.cu
CudaDomdec.h		CudaDomdec.h
CudaDomdecConstComm.cu		CudaDomdecConstComm.cu
CudaDomdecConstComm.h		CudaDomdecConstComm.h
CudaDomdecD2DComm.cu		CudaDomdecD2DComm.cu
CudaDomdecD2DComm.h		CudaDomdecD2DComm.h
CudaDomdecGroups.cu		CudaDomdecGroups.cu
CudaDomdecGroups.h		CudaDomdecGroups.h
CudaDomdecHomezone.cu		CudaDomdecHomezone.cu
CudaDomdecHomezone.h		CudaDomdecHomezone.h
CudaDomdecRecip.h		CudaDomdecRecip.h
CudaDomdecRecipComm.cu		CudaDomdecRecipComm.cu
CudaDomdecRecipComm.h		CudaDomdecRecipComm.h
CudaDomdecRecipLooper.cu		CudaDomdecRecipLooper.cu
CudaDomdecRecipLooper.h		CudaDomdecRecipLooper.h
CudaEnergyVirial.cu		CudaEnergyVirial.cu
CudaEnergyVirial.h		CudaEnergyVirial.h
CudaForcefield.h		CudaForcefield.h
CudaLeapfrogIntegrator.cu		CudaLeapfrogIntegrator.cu
CudaLeapfrogIntegrator.h		CudaLeapfrogIntegrator.h
CudaMPI.cpp		CudaMPI.cpp
CudaMPI.h		CudaMPI.h
CudaMonteCarloBarostat.cu		CudaMonteCarloBarostat.cu
CudaMonteCarloBarostat.h		CudaMonteCarloBarostat.h
CudaNeighborList.cu		CudaNeighborList.cu
CudaNeighborList.h		CudaNeighborList.h
CudaNeighborListBuild.cu		CudaNeighborListBuild.cu
CudaNeighborListBuild.h		CudaNeighborListBuild.h
CudaNeighborListSort.cu		CudaNeighborListSort.cu
CudaNeighborListSort.h		CudaNeighborListSort.h
CudaNeighborListStruct.h		CudaNeighborListStruct.h
CudaPMEDirectForce.cu		CudaPMEDirectForce.cu
CudaPMEDirectForce.h		CudaPMEDirectForce.h
CudaPMEDirectForceBlock.cu		CudaPMEDirectForceBlock.cu
CudaPMEDirectForceBlock.h		CudaPMEDirectForceBlock.h
CudaPMEForcefield.cu		CudaPMEForcefield.cu
CudaPMEForcefield.h		CudaPMEForcefield.h
CudaPMERecip.cu		CudaPMERecip.cu
CudaPMERecip.h		CudaPMERecip.h
CudaTopExcl.cu		CudaTopExcl.cu
CudaTopExcl.h		CudaTopExcl.h
Domdec.cpp		Domdec.cpp
Domdec.h		Domdec.h
DomdecD2DComm.cpp		DomdecD2DComm.cpp
DomdecD2DComm.h		DomdecD2DComm.h
DomdecGroups.cpp		DomdecGroups.cpp
DomdecGroups.h		DomdecGroups.h
DomdecMPI.cpp		DomdecMPI.cpp
DomdecMPI.h		DomdecMPI.h
DomdecRecip.h		DomdecRecip.h
DomdecRecipComm.cpp		DomdecRecipComm.cpp
DomdecRecipComm.h		DomdecRecipComm.h
DomdecRecipLooper.h		DomdecRecipLooper.h
EnergyVirial.cpp		EnergyVirial.cpp
EnergyVirial.h		EnergyVirial.h
Force.cu		Force.cu
Force.h		Force.h
Forcefield.h		Forcefield.h
HoloConst.cu		HoloConst.cu
HoloConst.h		HoloConst.h
KNOWN_ISSUES		KNOWN_ISSUES
LangevinPiston.cpp		LangevinPiston.cpp
LangevinPiston.h		LangevinPiston.h
LeapfrogIntegrator.h		LeapfrogIntegrator.h
Makefile		Makefile
Matrix3d.cu		Matrix3d.cu
Matrix3d.h		Matrix3d.h
MultiNodeMatrix3d.cpp		MultiNodeMatrix3d.cpp
MultiNodeMatrix3d.h		MultiNodeMatrix3d.h
README		README
ReduceList.h		ReduceList.h
XYZ.h		XYZ.h
XYZQ.cu		XYZQ.cu
XYZQ.h		XYZQ.h
cpu_transpose.cpp		cpu_transpose.cpp
cpu_utils.h		cpu_utils.h
cudaXYZ.h		cudaXYZ.h
cuda_utils.cu		cuda_utils.cu
cuda_utils.h		cuda_utils.h

Repository files navigation

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