pluto
pluto copied to clipboard
Memory usage optimisation
When compiling in release-small we should try and reduce the memory/RAM usage of the kernel. This would need a dynamic heap instead of the static 10% of available RAM heap. In addition, we can introduce memory profiling to get the maximum RAM used when running the runtime tests.
Memory profiling would be very nice to have. I wonder if we could add make an action that would profile a PR then compare that against a profile of the PR's base branch.
I reckon a dynamic heap size would be good to have for all build modes, but perhaps it would be better for performance if it was static on release-fast. We should look into it.