Using yarn.nodemanager.resource.detect-hardware-capabilities in Production

[#1]

I have a production Redhat HDP 3.1.5 cluster with different spec datanodes and cgroups enforced.

I am aware that through ambari, we can issue different configuration groups to different sets of nodemanagers to help us optimise the amount of resource allocated to yarn but can someone tell me if i can use yarn.nodemanager.resource.detect-hardware-capabilities instead? 

When configured would this enable yarn to auto-detect the hardware on a per node basis and save me having to use ambari's configuration groups?