Optimization Support #35
Description
It would be beneficial if we could automatically optimize the LLVM IR.
- Performance might improve.
- Inefficiencies in the IR get converted to inefficiencies in CIL, some of which cannot be fixed by Roslyn or the JIT compiler. For example, pinned memory allocations can only be optimized before conversion to CIL.
- Output size would likely be reduced.
- This is is particularly notable in regards to exception handling, where a lot of CIL instructions are generated for each LLVM IR instruction.
- Faster generation.
- Instruction count directly correlates with time spent generating.