Table 4.

Block matching running time (second) using CPU and GPU.

Case	Intel Intel(R) Xeon		Quadro 6000 GPU	AMD 6276		Tesla K20X GPU
	1 thread	12 threads		1 thread	16 threads
1	19.06	1.83	0.49	31.29	3.32	0.37
2	21.08	1.96	0.54	34.57	3.58	0.41
3	18.88	1.96	0.50	30.98	3.27	0.37
4	23.43	2.65	0.60	38.19	3.96	0.45
5	18.97	1.77	0.48	28.13	3.29	0.37
Average speedup		10.07	38.85		9.35	82.70

The speedup is with respect to 1 thread.