[llvm] [LLVM][DecoderEmitter] Add option to use function table in decodeToMCInst (PR #144814)

[Rahul Joshi via llvm-commits]

jurahul wrote:

> > ( 81.8%)  Two-Address instruction pass
> 
> This one, IIRC, scales linearly by the number of instructions. So it seems like the compilation speed slow down was primarily caused by the fact that the function is just too big.

There is still some mystery here since the # of instructions generated is the same in both cases (?). In the sense, the amount of code is the same, just split across functions (at a very coarse level). This to me indicates that either there is some code duplication happening in the switch-case version leading to super-linear increase in code size before two-address pass, or that the pass itself may be super-linear. But that would need more investigation with the actual test case. 

https://github.com/llvm/llvm-project/pull/144814