Homec4science
Diffusion lammps (efficient_neuronet) Browse

lammps/examples/intel/README
4a95b0047713efficient_neuronet

Last Change
Raw File

README
View Options

These example scripts can be run with the USER-INTEL package, assuming
you built LAMMPS with the package, either for CPU-only or with Intel
Xeon Phi coprocessor support.
If you run in another directory, please edit the read_data line to point to
the data.rhodo file. This is included in the LAMMPS distribution in the
'bench' directory.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Expected times are the expected LOOP times based on runs using dual-socket
Intel Xeon processor E5-2697 V2 with Intel Xeon Phi coprocessor 7120P.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
TO RUN WITHOUT A PACKAGE:
-----------------------------------------------------------------------------
mpirun -np 48 ../../src/lmp_YOUR_MACHINE -in in.intel.rhodo -log none -v b 0 -v s off
mpirun -np 48 ../../src/lmp_YOUR_MACHINE -in in.intel.lc -log none -v b 0 -v s off
TO RUN WITH OMP PACKAGE
-----------------------------------------------------------------------------
env OMP_NUM_THREADS=2 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.rhodo -log none -v b 0 -v s omp
+++++++++++++++++++++++
+ EXPECTED TIME: 20.33s
+++++++++++++++++++++++
env OMP_NUM_THREADS=1 mpirun -np 48 ../../src/lmp_YOUR_MACHINE -in in.intel.lc -log none -v b 0 -v s omp
+++++++++++++++++++++++
+ EXPECTED TIME: 19.92s
+++++++++++++++++++++++
TO RUN WITH INTEL+OMP PACKAGE WITHOUT OFFLOAD:
-----------------------------------------------------------------------------
env OMP_NUM_THREADS=2 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.rhodo -log none -v b 0 -v s intel
+++++++++++++++++++++++
+ EXPECTED TIME: 16.94s
+++++++++++++++++++++++
env OMP_NUM_THREADS=2 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.lc -log none -v b 0 -v s intel
+++++++++++++++++++++++
+ EXPECTED TIME: 5.80s
+++++++++++++++++++++++
TO RUN WITH INTEL+OMP PACKAGE WITH OFFLOAD TO XEON PHI (AUTO-BALANCED):
-----------------------------------------------------------------------------
env OMP_NUM_THREADS=1 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.rhodo -log none -v b -1 -v s intel
+++++++++++++++++++++++
+ EXPECTED TIME: 12.31s
+++++++++++++++++++++++
env OMP_NUM_THREADS=1 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.lc -log none -v b -1 -v s intel
+++++++++++++++++++++++
+ EXPECTED TIME: 4.00s
+++++++++++++++++++++++
TO RUN WITH INTEL+OMP PACKAGE WITH OFFLOAD TO XEON PHI (FIXED BALANCE):
-----------------------------------------------------------------------------
env OMP_NUM_THREADS=1 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.rhodo -log none -v b 0.68 -v s intel
+++++++++++++++++++++++
+ EXPECTED TIME: 11.40s
+++++++++++++++++++++++
env OMP_NUM_THREADS=1 mpirun -np 24 ../../src/lmp_YOUR_MACHINE -in in.intel.lc -log none -v b 0.53 -v s intel
+++++++++++++++++++++++
+ EXPECTED TIME: 3.93s
+++++++++++++++++++++++
c4science · Help