2d newtonian collission-less n-body simulation in c and cuda as of dec 24, 2014 added new parallel function for updating acceleration - using tiling and shared memory to hide global memory latency - 68x speed up on t10 on 2048 particles - 254x speed up on t10 on 4096 particles - 168x speed up on t10 on 8192 particles - 195x speed up on t10 on 16384 particles - 222x speed up on t10 on 32768 particles - 314x speed up on m2070 on 16384 particles added new parallel function that enhances data coalescence - 350x speed up on m2070 on 16384 particles added -use_fast_math flag in make file - 400x speed up on m2070 on 16384 particles - 292x speed up on t10 on 16384 particles as of dec 23, 2014 parallelized function for updating acceleration - 1 thread per particle, simple update - 31x speed up on t10 on 2048 particles - 52x speed up on t10 on 4096 particles - 75x speed up on t10 on 8192 particles - 62x speed up on t10 on 16384 particles as of dec 19, 2014 simple 2d sequantial collission less newtonian n-body simulation code - uniform spacial initialization with 0 initial velocity - particle-particle method (n^2), leapfrog update