@spatel, @david-arm: I integrated your proposed changes/updates and updated the patch. Local tests run fine, so it should be ready for committing from my side.

@spatel, @david-arm: Thanks for the update, I will integrate your proposed changes and update the patch after running some tests locally.

I reproduced the cause of the failure in the build-bot locally, fixed the bug and successfully ran the tests locally, so if you do not have any additional things that I should change, you could try again. Thanks!

@LukasSommerTu are you also looking into unifying/removing the code in SLPVectorizer/LV to create library calls?

I've found the reason for the build-bot failure, I had overlooked that codegen passes are currently pinned to the legacy PM in opt. I fixed this and also addressed @fhahn's inline comments (thanks for the feedback!).

The failure also seems to happen with X86 builds, so I suspect some connections to changes to the new PassManager, as the pass is not recognized at all. I will check that in detail and also make sure that the test works correctly, if the X86 target is not built.

@spatel: I got notified about some build-bot failures, so maybe it's better to revert the commit and I will check for the cause of the build-bot failures.

@spatel: Thanks for your quick feedback, I've addressed your comments and updated the patch.

Thanks @spatel for the detailed review!

I've updated the implementation to have this pass as an IR codegen pass.

I've updated the implementation to have this pass as an IR codegen pass.

Thanks to everyone for your feedback so far!

In D95373#2520441, @lebedev.ri wrote:

In D95373#2520409, @LukasSommerTu wrote:

@lebedev.ri: Thanks for your feedback!

My reasoning to implement this as an IR pass was that the replacement for the scalar version of the intrinsics is also happening as part of the middle-end, so it made sense to me to implement this in a similar location and use similar mechansisms.

Hm, it does? That seems quite surprising to me.

@lebedev.ri: Thanks for your feedback!

In D95373#2520302, @lebedev.ri wrote:

What's the intended use-case of this pass?

@jdoerfert: Can you please commit for me, I do not have commit access. Thanks!

Added a small test checking for functions with the correctly mangled name.

@jdoerfert: I've added a test. I kept it focused on the variants for NVPTX architecture (mangling is currently the same for all archs), let me know if that was what you had in mind.

@mehdi_amini: I do not have commit access, could you please commit the revision for me?

In D77643#1967140, @Kayjukh wrote:

I think it would make sense to propagate the changes to mlir-opt as well.

In the current design not. MlirOptMain does not receive the file itself, but rather an output stream, so it cannot manage the file.