Yes, one of the purposes of Conditions is to provide a path from the parent resource affected by some fault (especially one that is reflected in the HealthRollup) to the resource where the fault is actually described. We minimized the overhead by providing the Base Message Registry "ConditionExistsInResource" message so there's only one Condition entry per resource-that-needs-attention.
We would absolutely expect a DIMM fault to show up in the ComputerSystem Condition array, pointing to the particular DIMM resource.
I've proposed that we add another resource off of the Service Root to provide a place for a top-level rollup for everything that the service manages. We'd have to provide that as a separate resource, and not contained within the ServiceRoot schema because that is accessible without authentication.