[all-commits] [llvm/llvm-project] a51c2f: [SLP] no need to generate extract for in-tree uses...

[Enna1 via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: a51c2f39f5b6ba2cb03136016e707d2b8409eb0c
      https://github.com/llvm/llvm-project/commit/a51c2f39f5b6ba2cb03136016e707d2b8409eb0c
  Author: Enna1 <xumingjie.enna1 at bytedance.com>
  Date:   2023-12-30 (Sat, 30 Dec 2023)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
    M llvm/test/Transforms/SLPVectorizer/X86/extract_in_tree_user.ll
    M llvm/test/Transforms/SLPVectorizer/X86/reorder-reused-masked-gather2.ll

  Log Message:
  -----------
  [SLP] no need to generate extract for in-tree uses for original scala… (#76077)

…r instruction.

Before
https://github.com/llvm/llvm-project/commit/77a609b55636dc540090ef9105c60a99cfdbd1dd,
we always skip in-tree uses of the vectorized scalars in
`buildExternalUses()`,
that commit handles the case that if the in-tree use is scalar operand
in vectorized instruction,
we need to generate extract for these in-tree uses.

in-tree uses remain as scalar in vectorized instructions can be 3 cases:
- The pointer operand of vectorized LoadInst uses an in-tree scalar
- The pointer operand of vectorized StoreInst uses an in-tree scalar
- The scalar argument of vector form intrinsic uses an in-tree scalar

Generating extract for in-tree uses for vectorized instructions are
implemented in `BoUpSLP::vectorizeTree()`:
-
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp#L11497-L11506
-
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp#L11542-L11551
-
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp#L11657-L11667

However,
https://github.com/llvm/llvm-project/commit/77a609b55636dc540090ef9105c60a99cfdbd1dd
not only generates extract for vectorized instructions,
but also generates extract for original scalar instructions.
There is no need to generate extract for origin scalar instrutions,
as these scalar instructions will be replaced by vector instructions and
get erased later.

This patch marks there is no exact user for in-tree scalars that 
remain as scalar in vectorized instructions when building external uses,
In this case all uses of this scalar will be automatically replaced by extractelement.
and remove
-
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp#L11497-L11506
-
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp#L11542-L11551
-
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp#L11657-L11667
extracts.