SPEEDING UP XPDC1

1) Code can now be compiled with optimization (-O3) Flag ==> GAIN = 1.35

2) Time step can be increased by a factor of 3 while still keeping the number
   of missed collisions low and preserving the physics ==> GAIN = 3.0

3) Results of running Parallel XPDC1 for 1000 time steps on a cluster of
   3 Dual 400 MHz Pentium II machines connected by a 100 MB/s network.
   Input file is "case9.inp" (uniform mesh, 400K particles, initial Bessel
   profiles).       
                    #CPUs   Run-time (s)  Gain
                      1       506.8       1.00
                      2       252.2       2.01
                      3       172.3       2.94    ==> GAIN = (# CPUs)
                      4       134.4       3.77
                      5       106.6       4.75

4) Total GAIN = 1.35*3.0*(# CPUs) = 4.05*(# CPUs)