There was some talk of compile time memory management to help with GPU etc allocations.
In a similar vein: GitHub - doctorn/micro-mitten: You might not need your garbage collector
" Like Rust, micro-mitten offers a static approach to memory management; however, micro-mitten 's approach is significantly different from Rust’s. Rather than depending on single ownership and a complex lifetime system, micro-mitten uses a series of data-flow analyses to statically approximate heap liveness. This means that it maintains the ability to insert freeing code at appropriate program points, without putting restrictions on how you write your code. The theory behind the approach is documented in this thesis (Proust 2017).
Long story short, this is an attempt to see if we really can have unrestrictive compile-time garbage collection."
ccing @Keno in case anything in this repo and associated paper is helpful in that regard and posting here if anyone from the broader community has input.