CPU Utilization Graph for RDS instances is not marching what cloudwatch reports

A client reported high CPU utilization on one of their RDS instances, and upon checking on PMM we could see that the CPU utilization in PMM is nearly 0%.

The client shared the graph for the console and it indeed doesn't match. I checked on other environments and we have the same issue. 

I'm attaching 3 images:

• RDS Cloudwatch: It shows an Utilization from around 20-55%. 

• CPU Utilization PMM: It shows an utilization always less than 25%.

• PMM - CPU Utilization - TOP USAGE (OS Overview): Max usage of 45.60% which looks more accurate given the cloudwatch graphs.

Note that there is a time mismatch on those graphs because we added the node to PMM at 14:45 PST, but still it is enough evidence to give us an idea of the CPU usage reported at around that time.

I checked environments for other environments and the history is similar, there is data mismatch where PMM is reporting way less CPU utilization than CloudWatch.

In theory, the information should be the same as PMM collects metrics information from Cloudwatch, isn't it?

I tried changing the resolution to 5 mins as suggested on but the info still the data didn't match
 

There is concern that PMM is not showing reliable information.