Data randomization using SCV (a very power full mechanism for doing randomization).

1. In this site I show how to random simple fields in an ECC project.
   If more constraint driven complex scheme of randomization is required, one can use the SCV, which comes freely with systemc.


2. How to use SCV is well documented and can be easily found on WEB. In this example, however, I show to use it in a small c++ (cpp) project (not necessarily systemc project).


3. First a structure for the randomization target has to be declared.
   
   struct ecc_errors_t {
     unsigned char num_err; //number of errors
     unsigned char loc_err; //error array location
     unsigned char bit_err; //error bit location
   };
   


4. Next extend the SCV by the following:
   
   #ifndef DATA_EXT_H
   #define DATA_EXT_H
   
   #include "scv.h"
   #include "data.h"
   
   
   //declarations below this point are due to the --EXTEND_ALL option
   
   template<>
   class scv_extensions<ecc_errors_t> : public scv_extensions_base<ecc_errors_t> {
   public:
     scv_extensions<unsigned char > num_err;
     scv_extensions<unsigned char > loc_err;
     scv_extensions<unsigned char > bit_err;
   
     SCV_EXTENSIONS_CTOR(ecc_errors_t) {
       SCV_FIELD(num_err);
       SCV_FIELD(loc_err);
       SCV_FIELD(bit_err);
     }
   };
   
   #endif //DATA_EXT_H
   
   


5. Last we use it in a simple main cpp program:
   
   #include "systemc.h"
   #include "data_ext.h"
   
   
   int main (int argc, char *argv[]) {
   
     if(argc > 1) {
       scv_random::set_global_seed(atoi(argv[1]));
     }
   
     scv_smart_ptr<ecc_errors_t> data_p("My_data");
     data_p->num_err.keep_only(0, 2);
     data_p->loc_err.keep_only(5, 255);
     data_p->bit_err.keep_only(0, 7);
   
     data_p->next();
     cout <<
     "num=" << (unsigned)data_p->num_err << endl;
     cout <<
     "loc=" << (unsigned)data_p->loc_err << endl;
     cout <<
     "bit=" << (unsigned)data_p->bit_err << endl;
     return 0;
   
   }
   


6. I use the make file, which is given in the SCV examples:
   make -f Makefile.osci
   TOPDIR=/mnt/Home/pini/Home_1/pini/systemc_2.2.0/examples/scv
   
   SOURCE_FILES=scv_rand_1.cpp
   include ${TOPDIR}/Makefile.rules
   
   This creates an executable named run.x and runs it. To run it, without the makefile (./run.x), you need to define the following variable:
   
   LD_LIBRARY_PATH=.:/mnt/Home/pini/Home_1/pini/systemc_2.2.0/lib-linux:/mnt/Home/pini/Home_1/pini/systemc_2.2.0/lib-linux export LD_LIBRARY_PATH



7. Recently I tried to use systemc and scv on linux machine with cadence tools. I wanted a little, standalone program, which randomize some numbers with the simple constrains as scv allows. I did not have much time to spend on the best way do it with cadence installation, so I came out with the following script to compile and link the program. This is not ideal, but worked well for me:
   
   
   1. #!/bin/bash
   3. echo $LD_LIBRARY_PATH | grep cadence
   4. if [ $? -ne 0 ] ; then
   5.   setenv LD_LIBRARY_PATH ${LD_LIBRARY_PATH}:/tools/cadence/INCISIV12.2-s005/tools.lnx86/lib/64bit
   6. fi
   7. rm -f a.out *.o
   8. clear
   9. g++ -I/tools/cadence/INCISIV12.2-s005/tools.lnx86/systemc/include -I/tools/cadence/INCISIV12.2-s005/tools.lnx86/tbsc/include main.cpp -ldl -L/tools/cadence/INCISIV12.2-s005/tools.lnx86/lib/64bit -lrnc /tools/oss/packages/x86_64-rhel4/gcc/4.1.2/lib64/libstdc++.a /tools/cadence/INCISIV12.2-s005/tools.lnx86/systemc/lib/64bit/libsystemc_ar.a /tools/cadence/INCISIV12.2-s005/tools.lnx86/tbsc/lib/64bit/libscv.a
   10. echo $LD_LIBRARY_PATH