Constraints vs keyframes

we are developing content for rural schools in South America, where many users access our applications on low-end devices with limited performance capabilities.

In scenarios where I have multiple elements that need to rotate, would it be more efficient to use one controller with rotation constraints linked to each element, or would it be better to key each element individually for better performance on these lower-end devices?

Really appreciate your help in figuring out the best way to keep things running smooth and efficient!