An Evaluation Model and Benchmark for Parallel Computing Frameworks

<table class="table-group" id="tab5"><tr><td><table class="table"><tr><td class="thead-hr" colspan="6"><hr/></td></tr><tr class="thead"><td class="align_left">Input</td><td class="align_left">10G</td><td class="align_left">20G</td><td class="align_left">30G</td><td class="align_left">40G</td><td class="align_left">50G</td></tr><tr><td class="thead-hr" colspan="6"><hr/></td></tr><tr><td class="align_left">Spark</td><td class="align_left">1240</td><td class="align_left">2041</td><td class="align_left">2482</td><td class="align_left">3620</td><td class="align_left">5218</td></tr><tr><td class="align_left">MARS</td><td class="align_left">1511</td><td class="align_left">2315</td><td class="align_left">2913</td><td class="align_left">3110</td><td class="align_left">4258</td></tr><tr><td class="align_left">Speedup</td><td class="align_left">1.21</td><td class="align_left">1.13</td><td class="align_left">1.17</td><td class="align_left">0.85</td><td class="align_left">0.81</td></tr><tr class="table-tr"><td colspan="6"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

<div>Spark’s speedup over MARS on running logistic regression.</div>

Mobile Information Systems

tab5

Table 5

Table 5: An Evaluation Model and Benchmark for Parallel Computing Frameworks 