llvm dev - Jul 2020 - How to get information about data dependencies?

Hello,

I had a question similar to this one in the link below.

http://lists.llvm.org/pipermail/llvm-dev/2015-January/080589.html

In addition, if we can't get RAW dependencies information using the
provided llvm tools, how would I write a pass for it? Are there any helpful
resources?

Thanks,
Jiho
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20200701/78f95e90/attachment.html>

LLVM has multiple dependence analyses, each with its up- and downsides:

 * llvm::DependenceAnalysis
 * llvm::LoopAccessAnalysis
 * llvm::MemorySSA
 * polly::DependenceInfo

Just running the analysis passes does not change anything, the
analysis has to be used by a transformation pass. Since there is no
common interface for dependence analysis, each transformation pass
expects a specific analysis and runs it itself if necessary. For
instance LoopVectorize uses LoopAccessAnalysis.

None of these are well documented(*), I suggest to look at the
transformation pass sources that use them.

Michael

(*) For MemorySSA, there is https://www.llvm.org/docs/MemorySSA.html

Am Mi., 1. Juli 2020 um 19:52 Uhr schrieb Jiho Ro via llvm-dev
<llvm-dev at lists.llvm.org>:
>
> Hello,
>
> I had a question similar to this one in the link below.
>
> http://lists.llvm.org/pipermail/llvm-dev/2015-January/080589.html
>
> In addition, if we can't get RAW dependencies information using the
provided llvm tools, how would I write a pass for it? Are there any helpful
resources?
>
> Thanks,
> Jiho
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev

Michael Kruse via llvm-dev <llvm-dev at lists.llvm.org> writes:
> LLVM has multiple dependence analyses, each with its up- and downsides:
>
>  * llvm::DependenceAnalysis
>  * llvm::LoopAccessAnalysis
Can someone explain the differences between these?  As far as I can tell
they essentially do the same thing (though perhaps one is more
precise?).  LAA seems to be used by vectorization (what else?) while DA
seems to be used by loop transformations (what else?).

I am not a loop opt guy so while I'm familiar with the basic ideas, the
details are somewhat lost on me.  Is there a reason to have two passes
or should they be combined and maintained as one pass?  If I have need
of dependence analysis it's not clear which I should use and the
comments are not much help.

                   -David