Compute units used in (re)deployment actions

Is there a rule of thumb to how many compute units are used each time I re-deploy an auto-scale project? Or is there a way to see how many compute units are used for a re-deployment action and requests separately?

I am trying to extrapolate how much an autoscale deployment may cost based on requests as I test my application. When I look at the compute units usage data I am unsure how much is from requests vs my frequent re-deployments in this testing stage.