[Mlir-commits] [mlir] [mlir] [memref] Compile-time memref.alloc Scheduling/Merging optimization (PR #95882)

[llvmlistbot at llvm.org]

https://github.com/MaheshRavishankar commented:

This is really cool and thanks a lot for the writeup! This is a large PR, but seems structured enough. Overall the idea of using a linear scan allocator makes sense and also the region based allocation makes sense. There are a few details that I need to look at the implementation to get better understanding of (like how do you manage to limit/recover gaps within the allocation, etc.). 
It might take a while, but I am happy to review this.

https://github.com/llvm/llvm-project/pull/95882