So, um, I would like to look at your patches, but I'm not having much success running "arc patch" against them.

@SamB, this patch is highly out of sync with the upstream and won't work in its current state. The VPlan infrastructure has significantly changed since this patch. If you want to focus on the VPlan part of this patch, patch stack - D104608, D104610, D104673, D104979 is a bit more recent although I suspect they would still conflict with the upstream.

Rebasing on more recent upstream main.

@simoll I believe there is a way to commit without changing the author, such that both phabricator and github show something like "authored by <author> and committed by <committer>".

Can someone please commit it on my behalf? I don't have commit access.

addressed typos found by @frasercrmck.

Ping!

Ping!

Just a friendly ping!

Addressed comments.

In D103898#2807243, @simoll wrote:

As discussed on the call, maybe rename llvm.experimental.vp.vector.splice to llvm.experimental.vp.splice.. the "vector" is implied

In D103898#2812522, @simoll wrote:

This patch needs to implement VPIntrinsic::getDeclarationForParams.

Addressed @simoll's comment. Took the liberty to clang-format part of the test for memory ops.

Address VPlan related comments.

In D103898#2809074, @craig.topper wrote:

Are evl1 and evl2 likely to be the same value? Do you have an example when they would be different?

LGTM.

Thanks @fhahn for a look at the VPlan bits of the patch.

LGTM.

Thanks @sdesmalen for the patch!
Just a couple of clarifying questions.
Also, all the non-target specific test cases only check for -scalable-vectorization=on and with metadata hint enabled (except for one that focuses on width hint). Just for the sake of completeness, more tests with different values for the flag and its combination with the metadata hint would be nice to have.

Sync up with ongoing changes in VPlan. Minor refactor.

In D100941#2716539, @rogfer01 wrote:

@vkmr do you have commit rights? I can commit this for you.

Addressed review comments by @craig.topper and @frasercrmck.

@simoll thanks for putting this up. Wondering if this would be the right place to add a link to the discord channel too!

Merged more recent changes on main.

Revert diff 334860

Remove unused header

Remove unused header include

Thanks Hussain for this patch!
For now I have a couple of minor comments, I will take a deeper look later.
Also, I don't quite understand yet how VP handles (or plans to handle) specifying alignment parameter. Does it need VPBuilder (not to be confused with VPlan VPBuilder!) for that? Would it make sense to add the Alignment parameter to VP load/store/scatter/gather intrinsics (much like their masked counterparts) for now, unless there is already a way to specify that?

FWIW, we have a similar intrinsic in our downstream compiler for RISC-V. We call it experimental.vector.slideleftfill and is same as flavour 2 of this (except the "offset" need not be an immediate) as suggested by @sdesmalen. We have a separate predicated version (experimental.vector.vp.slideleftfill) too which takes additional arguments for the explicit vector length of the two vectors.

LGTM.

Other than the minor nit, LGTM.

Overall, definitely a good starting point. I just have a couple of minor comments.

Thanks @fpetrogalli for this proposal! I agree with @rogfer01 and @rengolin in that this approach makes a lot of sense.
I have added a few comments inline. Overall looks good.