For Java and Kotlin, multiple submits produce variances in both runtime and memory. But there seems to be two separate groupings of runtime and memory. When memory usage is noticeably higher, then runtime is also higher. When memory usage is lower, then runtime is also lower. Within these high and low groupings, the runtime and memory values vary between different submits, but these high and low groupings seem to remain distinct and vary near their own median values. This has been true for many leetcode problems.

These runtime and memory usage groupings are almost as though the leetcode problem is being compiled and executed on two different types of servers. One server where code runtimes are faster and requires less memory. The other server where code runtimes are slower and require more memory. Or maybe it is a difference in task priority.

Does anybody know why there are these two groupings of runtime and memory usage values?

The graphs below are scatter plots of two Leetcode problems' Runtime vs. Memory Usage. For some reason, the second graph contains 3 separate groupings.