accelerInt
v0.1
|
The following options can be passed to SCons to customize the accelerInt build process. They should be given in the form:
scons option1=value1 option2=value2
Variables set in this way will be stored in the accelerInt.conf
file and reused automatically on subsequent invocations of scons. Alternatively, the configuration options can be entered directly into accelerInt.conf
before running scons build
. The format of this file is:
option1 = 'value1' option2 = 'value2'
To build only the CPU solvers, SCons can be called as:
scons cpu
Similarly, the GPU solvers can be built with:
scons gpu
toolchain | [ string ] The C/C++ compilers to use. - default: 'gnu' |
blas_lapack_libs | [ string ] Comma separated list of blas/lapack libraries to use for the various solvers, set blas_lapack_libs to the the list of libraries that should be passed to the linker, separated by commas, e.g. "lapack,blas" or "lapack,f77blas,cblas,atlas". - default: 'mkl_rt,mkl_intel_lp64,mkl_core,mkl_gnu_thread,dl,mkl_mc,mkl_def' |
blas_lapack_dir | [ /path/to/blas_lapack_dir ] Directory containing the libraries specified by 'blas_lapack_libs'. - default: 'usr/local/lib' |
NVCCFLAGS | [ string ] Compiler flags passed to the CUDA compiler, regardless of optimization level. - default: '-m64 -Xptxas -v' |
CCFLAGS | [ string ] Compiler flags passed to both the C and C++ compiler, regardless of optimization level - default: '-m64' |
CXXFLAGS | [ string ] Compiler flags passed to only the C++ compiler, regardless of optimization level - default: '' |
CFLAGS | [ string ] Compiler flags passed to only the C compiler, regardless of optimization level - default: '-std=c99' |
thread_flags | [ string ] Compiler and linker flags for POSIX multithreading support. - default: '-pthread' |
openmp_flags | [ string ] Compiler and linker flags for OpenMP support. - default: '-fopenmp' |
compute_level | [ string ] The CUDA compute level of your GPUs - default: 'sm_20' |
sundials_inc_dir | [ string ] The directory where the sundials headers are located - default: 'usr/local/include' |
sundials_lib_dir | [ string ] The directory where the sundials libraries are located - default: 'usr/local/lib' |
boost_inc_dir | [ string ] The directory where the boost headers are located - default: 'usr/local/include' |
fftw3_inc_dir | [ string ] The directory where the FFTW3 headers are located - default: 'usr/local/include' |
fftw3_lib_dir | [ string ] The directory where the FFTW3 libraries are located - default: 'usr/local/lib' |
mechanism_dir | [ string ] The directory where mechanism files are located. - default: '/home/nick/Dropbox/accelerInt/mechanism' |
DEBUG | [ yes | no ] Compiles with Debugging flags and information. - default: 'no' |
ATOL | [ string ] Absolute Tolerance for integrators - default: '1e-10' |
RTOL | [ string ] Relative Tolerance for integrators - default: '1e-6' |
t_step | [ string ] Step size for integrator - default: '1e-6' |
t_end | [ string ] End time of the integrator - default: '1e-6' |
N_RA | [ string ] The size of the Rational Approximant for the Exponential Integrators. - default: '10' |
SAME_IC | [ yes | no ] Use the same initial conditions (specified during mechanism creation) during integration. - default: 'no' |
SHUFFLE | [ yes | no ] Shuffle the PaSR initial conditions. - default: 'no' |
PRECONDITION | [ yes | no ] Precondition (via clustering) the PaSR initial conditions. - default: 'no' |
[ yes | no ] Log output to screen. - default: 'no' | |
LOG_OUTPUT | [ yes | no ] Log output to file. - default: 'no' |
LOG_END_ONLY | [ yes | no ] Log only beginning and end states to file. - default: 'no' |
IGN | [ yes | no ] Log ignition time. - default: 'no' |
FAST_MATH | [ yes | no ] Compile with Fast Math. - default: 'no' |
FINITE_DIFFERENCE | [ yes | no ] Use a finite difference Jacobian (not recommended) - default: 'no' |
DIVERGENCE_WARPS | [ string ] If specified, measure divergence in that many warps - default: '0' |
CV_HMAX | [ string ] If specified, the maximum stepsize for CVode - default: '0' |
CV_MAX_STEPS | [ string ] If specified, the maximum stepsize for CVode - default: '20000' |
CONST_TIME_STEP | [ for logging purposes ] If specified, adaptive timestepping will be turned off - default: 'no' |