https://community.intel.com/t5/Software-Tuning-Performance/Intel-MBM-per-NUMA-node-bandwidth-usage-info/m-p/1158167#M6986 <P>I don't think you can measure bandwidth per NUMA node using MBM because there is only one Event ID (0x02) for the total external bandwidth and Event ID (0x3) for the local memory bandwidth. However, there is one PMON box per QPI/UPI link. You can use the data flit events to measure data QPI bandwidth between any two specific sockets as follows:</P><P>Incoming data bandwidth = (RxL_FLITS_G1.DRS_DATA + RxL_FLITS_G2.NCB_DATA) * 8 / time</P><P>Outgoing data bandwidth = TxL_FLITS_G0.DATA * 8 / time</P> Thu, 28 Feb 2019 20:03:19 GMT HadiBrais 2019-02-28T20:03:19Z https://community.intel.com/t5/Software-Tuning-Performance/Intel-MBM-per-NUMA-node-bandwidth-usage-info/m-p/1158166#M6985 <P>Does Memory Bandwidth Monitoring technology collect "per numa&nbsp;node" bandwidth usage info or it's just local and all the remote nodes?</P> Sun, 24 Feb 2019 05:30:44 GMT https://community.intel.com/t5/Software-Tuning-Performance/Intel-MBM-per-NUMA-node-bandwidth-usage-info/m-p/1158166#M6985 Reza1 2019-02-24T05:30:44Z https://community.intel.com/t5/Software-Tuning-Performance/Intel-MBM-per-NUMA-node-bandwidth-usage-info/m-p/1158167#M6986 <P>I don't think you can measure bandwidth per NUMA node using MBM because there is only one Event ID (0x02) for the total external bandwidth and Event ID (0x3) for the local memory bandwidth. However, there is one PMON box per QPI/UPI link. You can use the data flit events to measure data QPI bandwidth between any two specific sockets as follows:</P><P>Incoming data bandwidth = (RxL_FLITS_G1.DRS_DATA + RxL_FLITS_G2.NCB_DATA) * 8 / time</P><P>Outgoing data bandwidth = TxL_FLITS_G0.DATA * 8 / time</P> Thu, 28 Feb 2019 20:03:19 GMT https://community.intel.com/t5/Software-Tuning-Performance/Intel-MBM-per-NUMA-node-bandwidth-usage-info/m-p/1158167#M6986 HadiBrais 2019-02-28T20:03:19Z