[llvm] [DependenceAnalysis] Extending SIV to handle fusable loops (PR #128782)

[via llvm-commits]

amehsan wrote:

> > I believe I have mentioned this multiple times, but the key point here is that it is checked that the two inner loops have the same backedge-taken count. How this value is calculated is irrelevant.
> 
> I apologize if I made you feel like you had to explain the same thing multiple times. However, your last comment made me realize the following new points.
> 
> > How this value is calculated is irrelevant.
> 
> This is exactly the part where I cannot be confident. I'm not sure whether it really doesn’t matter how the backedge-taken counts are computed. But maybe I'm just overthinking (which is why I called myself "nitpicky" in the previous comment), and this might actually be irrelevant.
> 
> So, sorry to have made you explain at such length, but I _personally_ still cannot feel fully confident about correctness. That said, I’ll refrain from further comments on correctness. If someone else reviews this discussion and believes it is correct, they will probably approve this PR. Thanks again for taking the time to explain it.

Thank you and I appreciate your attention to the quality of the open source compiler.

https://github.com/llvm/llvm-project/pull/128782