Trends will always put all objects in the same graph. selecting only 1 in the config will just simply show you the first VM not generate different trends with one VM in each.
The only real way to do this is in 6.7 and later you can create a dashboard with a object list showing a group with all the VMs in it and a metric chart. selecting the group it will show you all the VMs cpu usage in separate charts.
The problem with that is dashboards never format well into reports but in 7.0 you can send out a never expiring no login link to the dashboard instead of running and sending out reports.
Just be aware of what has changed in 7.0 before/if you upgrade