[llvm] [SLP] Avoid seeding related affine loop address computations (PR #226220)

Alexey Bataev via llvm-commits llvm-commits at lists.llvm.org
Sun Oct 11 09:48:52 PDT 2026


================
@@ -29418,6 +29419,11 @@ bool SLPVectorizerPass::runImpl(Function &F, ScalarEvolution *SE_,
     // Start new block - clear the list of reduction roots.
     R.clearReductionData();
     collectSeedInstructions(BB);
+    // Saved to keep their index computations out of the once-used seed attempt.
+    if (VectorizeOnceUsed)
+      for (const GEPList &List : make_second_range(GEPs))
+        if (List.size() >= 2)
+          append_range(IndexedGEPs[BB], List);
----------------
alexey-bataev wrote:

```suggestion
```

https://github.com/llvm/llvm-project/pull/226220


More information about the llvm-commits mailing list