case 1:
-------

	1(a): No dependence exists.

	1(b): Parallelization is done by gcc-4.7.2 and can be verified
	      by looking at parallelization dump.
	      
	     -----------------------------------------------------
	      D.1896_6 = __builtin_omp_get_num_threads ();
  	      D.1897_7 = (<unnamed-unsigned:32>) D.1896_6;
 	      D.1898_8 = __builtin_omp_get_thread_num ();
      	      D.1899_9 = (<unnamed-unsigned:32>) D.1898_8;
	     -----------------------------------------------------

	      Similarly Vectorization is done by gcc-4.7.2, and can be
	      verified by looking at vectorization dumps.

	     -----------------------------------------------------
	      # vect_pb.5_6 = PHI <vect_pb.5_13(4), vect_pb.8_10(2)>
 	      # vect_pa.10_16 = PHI <vect_pa.10_17(4), vect_pa.13_15(2)>
	      vect_var_.9_14 = MEM[(int[5000000] *)vect_pb.5_6];
	      vect_pb.5_13 = vect_pb.5_6 + 16;
	      vect_pa.10_17 = vect_pa.10_16 + 16;
	     -----------------------------------------------------

	 2  : Execution time may differ for different machines but
	      they are expected to be as follows.
	      ET(upuv) > ET(v) and ET(upuv) > ET(p)
	
	      where ET(upuv)  is execution time of unparallelized
	      and unvectorized code, ET(p)  is execution time of 
	      parallelized  code and  ET(v) is  execution time of 
	      vectorized code.

--------------------------------------------------------------------------------

Case 2:
-------

	1(a): In case of array 'a' write precedes read option,
	      Array reference    :   a				
	      Type of dependence :   RAW (loop carried)		
	      Source  		 :   S1	
	      Sink		 :   S2	
	     ------------------------------------------------------
	      Array reference    :   b				
	      Type of dependence :   RAW (loop carried)		
	      Source  		 :   S2	
	      Sink		 :   S1	

	1(b): Program is  neither  vectorizable nor parallelizable
	      this can easily be verified by looking  at the  dumps


	 2  : Since gcc-4.7.2 can not parallelize or vectorize the code,
	      therefore ET(upuv) ET(v) and ET(p) are almost equal

	      where ET(upuv)  is execution time of unparallelized
	      and unvectorized code, ET(p)  is execution time of 
	      parallelized  code and  ET(v) is  execution time of 
	      vectorized code.