HPC Porject
OpenMP parallelization by Sizhang Dai
[email protected]

How to compile
use "make".

All the files or functions' names with '_para' suffix are paralleized using OpenMP
All the files that I made add or change to original libaray are listed in 'Modified File Tree.txt'


The 'testing_para' folder contains testing code for corretness and porformance tests
testing_para/example contains performance testing code for vector and matrix functions as described below
testing_para/exConv1d_f contains testing code for main function in 1D convolution_f module
testing_para/exCorr1d_f contains testing code for main function in 1D correlation_f module
usage: 'make' in one of the folders

use enviroment variables OMP_NUM_THREADS to set the number of threads will be used: 
export OMP_NUM_THREADS=X (X is the number of threads you want to use)


modules functions:
convolve1d_f: 
            vsip_convolve1d_f_para()
        inlcuding sub-functions the will be used in the vsip_convolve1d_f_para():
                vsip_vadd_f_para
                vsip_vcopy_f_f_para
                vsip_cvcopy_f_f_para
                vsip_cvconj_f_para
                vsip_rscvmul_f_para
                vsip_svmul_f_para
                vsip_cvmul_f_para
                VI_sort_copy_f_para
                VI_dft_f_para
                VI_ccfftip_f_para
                VI_ccfftmip_f_para
                vsip_ccfftip_f_para
                vsip_ccfftmip_f_para
                vsip_cvmmul_f_para
                VI_sortm_copy_f_para


correlate1d_f: 
                vsip_correlate1d_f_para()  for real number
            &&  vsip_ccorrelate1d_f_para() for complex number


matrix functions:
            vsip_gems_f_para, 
            vsip_gemp_f_para,


vector functions:
            vsip_vsumval_f_para, 
            vsip_vsumsqval_f_para, 
            vsip_vdot_f_para, 
            vsip_vmul_f_para,