Just looking to see if anyone else has estimated (and how successful was the estimate) on using data from provision redshift cluster to estimate redshift serverless cost

So for example if you have a single node dc2.large (rough equal to 1 rpu) and average usage of 6 hours of query excecution (according to beloq query not 100% sure if this is the correct?)

SELECT 
    date_trunc('day', starttime) AS query_date,
    ((SUM(EXTRACT(EPOCH FROM (endtime - starttime)))/60)/60) AS total_query_time_hours
FROM stl_query
WHERE userid > 1
  AND starttime >= date_trunc('month', current_date)
GROUP BY 1
ORDER BY 1 DESC;

So this states you are using 6 hrs average per day. Looking at minimums of base 8 RPUs on serverless, assuming no performance/speed improvements & no autoscaling from the 8 RPUs on existing workloads, is it as simple as 6 (hours active) * 8 (number of RPUs) * 0.36 (current pricing)? $17.28 per day?

Do all 8 "base capacity" get used on every query or does it depend on complexity of queries etc?