[llvm] [LV][VPlan] Add initial support for CSA vectorization (PR #106560)

[Michael Maitland via llvm-commits]

================
@@ -0,0 +1,5188 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -S -mtriple riscv64 -mattr="+v" -riscv-v-vector-bits-min=256 \
+; RUN:   -passes=loop-vectorize -force-tail-folding-style=data-with-evl \
+; RUN:   -enable-csa-vectorization | FileCheck %s -check-prefix=EVL
+; RUN: opt < %s -S -mtriple riscv64 -mattr="+v" -riscv-v-vector-bits-min=256 \
+; RUN:   -passes=loop-vectorize -force-tail-folding-style=none \
+; RUN:   -enable-csa-vectorization | FileCheck %s -check-prefix=NO-EVL
+; RUN: opt < %s -S -mtriple riscv64 -mattr="+v" -riscv-v-vector-bits-min=256 \
+; RUN:   -passes=loop-vectorize -force-tail-folding-style=data \
+; RUN:   -enable-csa-vectorization | FileCheck %s -check-prefix=DATA
----------------
michaelmaitland wrote:

We are generating different recipes for different tail folding styles. We should test against tail folding styles since that gives us coverage for those code paths.

https://github.com/llvm/llvm-project/pull/106560