[PATCH] D101290: [LV] Try to sink and hoist inside candidate loops for vectorization.

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri accepted this revision.
lebedev.ri added a comment.
This revision is now accepted and ready to land.

Cautious LG, you likely want one more review.



================
Comment at: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:9998
+    // Remove block if it became dead.
+    if (pred_begin(BB) == pred_end(BB)) {
+      LI.removeBlock(BB);
----------------
pred_empty(BB)


================
Comment at: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:10008-10011
+        BI->getSuccessor(0)->getSinglePredecessor() &&
+        BI->getSuccessor(1)->getSinglePredecessor() &&
+        all_of(successors(BB), [&LI, L](BasicBlock *Succ) {
+          return LI.getLoopFor(Succ) == L;
----------------



================
Comment at: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:10015
+      auto *OldElse = BI->getSuccessor(1);
+      if (hoistThenElseCodeToIf(BI, TTI, false, &DTU)) {
+        Changed = true;
----------------



================
Comment at: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp:10006-10016
+      if (hoistThenElseCodeToIf(BI, TTI, false, &DTU)) {
+        Changed = true;
+        SmallPtrSet<BasicBlock *, 2> NewSuccs(succ_begin(BB), succ_end(BB));
+        if (!NewSuccs.count(OldThen)) {
+          LI.removeBlock(OldThen);
+          DeleteDeadBlock(OldThen);
+        }
----------------
fhahn wrote:
> lebedev.ri wrote:
> > I think it would be fine to sink that into `hoistThenElseCodeToIf()`.
> Unfortunately I think that would be incompatible with the use in SimplifyCFG, which does not seem to handle removing BBs other than the current one well.
> 
> But I adjusted the loop to remove any blocks that became dead at the top-level, which hopefully makes things a bit clearer.
Hmm, true. Disregard then.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D101290/new/

https://reviews.llvm.org/D101290