NUMA-Aware Thread Scheduling for Big Data Transfers over Terabits Network Infrastructure

<table class="fixed-width table-group" id="tab2"><tr><td><table class="table"><colgroup><col style="width:17.03em"/><col style="width:9.18em"/></colgroup><tr><td class="thead-hr" colspan="2"><hr/></td></tr><tr class="thead"><td class="align_left"> Thread placement </td><td class="align_center"> Throughput (MB/s) </td></tr><tr><td class="thead-hr" colspan="2"><hr/></td></tr><tr><td class="align_left">All threads placed on same CPU socket</td><td class="align_center"> 2065.21</td></tr><tr><td class="align_left">CT placed on different CPU socket</td><td class="align_center"> 1914.02</td></tr><tr><td class="align_left">MT placed on different CPU socket</td><td class="align_center"> 1719.80</td></tr><tr class="table-tr"><td colspan="2"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

<div>The evaluation of thread placement across CPU sockets. Total 8 I/O threads are used in this experiment.</div>

Scientific Programming

tab2

Table 2

Table 2: NUMA-Aware Thread Scheduling for Big Data Transfers over Terabits Network Infrastructure 