Kubernetes workload allocated compute cost

I'm trying to come up with a semi-reliable formula for workload cost (i.e. on-demand compute cost)

Here's a sample example :

• Running a k8s cluster with 3 aws m5.xlarge nodes (4vCPU/16GiB - 32$/week per node)
• There's a workload (simple deployment) requesting 2.5vCPU and 8GiB Mem overall (utilization under request)

Is this the correct calculation to determine the compute cost for the workload?

Max(per unit CPU Cost, per Unit Mem Cost) = Max([2.5/4]*32, [8/16]*32) = Max(20,16)=$20/week

Moreover, Should we consider how the Infra is handling the overall workloads as well ? (e.g. factor in the cluster Request/Capacity ratio as well)

I've already looked at the OpenCost (Opensourced Kubecost) but wondering what other tool/reference/formula is usually used usually used for this topic