An Evaluation Model and Benchmark for Parallel Computing Frameworks

Table 9

Spark’s speedup over MARS on running the Bayesian classification.

Input	10G	20G	30G	40G	50G
Spark	1021	1564	1971	2677	4923
MARS	851	932	992	2128	4491
Speedup	0.83	0.60	0.50	0.79	0.91