Installation: Compiler and Flags

To choose a compiler and compilation flags, set the following variables in the Makefile:

CXX        =    # C++ compiler
CXXFLAG    =    # compilation flags
OPENMPFLAG =    # openmp flag
LIB        =    # libraries and linker flags
NVCC       =    # CUDA compiler

Example: Intel compiler

CXX        = $(MPI_PATH)/bin/mpicxx   # replace by "icpc" in the serial mode
CXXFLAG    = -g -O3 -w1
OPENMPFLAG = -fopenmp
LIB        = -limf
NVCC       = $(CUDA_PATH)/bin/nvcc

Example: GNU compiler

CXX        = $(MPI_PATH)/bin/mpicxx   # replace by "g++" in the serial mode
CXXFLAG    = -g -O3 -Wall -Wextra
CXXFLAG   += -Wno-unused-variable -Wno-unused-parameter \
             -Wno-maybe-uninitialized -Wno-unused-but-set-variable \
             -Wno-unused-result -Wno-unused-function
OPENMPFLAG = -fopenmp
LIB        =
NVCC       = $(CUDA_PATH)/bin/nvcc

On a Cray system (even when adopting the Intel or GNU compiler), set CXX and NVCC as

CXX        = CC
NVCC       = nvcc -ccbin CC

The prefixes $(MPI_PATH)/bin/ and $(CUDA_PATH)/bin/ in the above examples are optional to force the Makefile to use the correct compiler, where MPI_PATH and CUDA_PATH are the library paths described in External Libraries.

Links

• Makefile configuration -- Simulation Options
• Makefile configuration -- External Libraries
• Back to the main page of Installation