[llvm] [SLP]Initial support for (masked)loads + compress and (masked)interleaved (PR #132099)
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Fri Mar 21 04:46:56 PDT 2025
================
@@ -3375,10 +3377,11 @@ class BoUpSLP {
/// (either with vector instruction or with scatter/gather
/// intrinsics for store/load)?
enum EntryState {
- Vectorize, ///< The node is regularly vectorized.
- ScatterVectorize, ///< Masked scatter/gather node.
- StridedVectorize, ///< Strided loads (and stores)
- NeedToGather, ///< Gather/buildvector node.
+ Vectorize, ///< The node is regularly vectorized.
+ ScatterVectorize, ///< Masked scatter/gather node.
+ StridedVectorize, ///< Strided loads (and stores)
+ MaskedLoadCompressVectorize, ///< Masked load with compress.
----------------
RKSimon wrote:
Aren't all compress nodes masked? Would CompressedVectorize be better?
https://github.com/llvm/llvm-project/pull/132099
More information about the llvm-commits
mailing list