[PATCH] D125287: [SLP] Improve root steering by building actual trees instead of calling the look-ahead heuristic

Tue May 10 09:03:50 PDT 2022

vdmitrie added inline comments.

================
Comment at: llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp:2028-2036
+      SmallVector<Value *, 2> Roots(
+          {Candidates[I].first, Candidates[I].second});
+      buildTree(Roots, /*UserIgnoreList=*/None,
+                RootLookAheadMaxSize.getValue());
+      if (isTreeTinyAndNotFullyVectorizable())
+        continue;
+      reorderTopToBottom();
----------------
ABataev wrote:
> vporpo wrote:
> > vdmitrie wrote:
> > > ABataev wrote:
> > > > I'm afraid of increasing compile time. All this stuff includes scheduling, which may take lots of time for large basic blocks.
> > > Yep, I agree. this will be more expensive for compile time.
> > > What about combining both worlds? I mean first try to use lookahead heuristics to get the single best. And if we can't narrow down to just one pair only then switch into probing via building trees.
> > > I believe it will not happen too frequently. We can also increase lookahead depth to make it even less frequent when we need to build vectorizable tree.
> > What if we set a flag to disable scheduling for these types of fast tree estimations? 
> There is a problem with this fix that it tries to avoid/mask the problem, not fix it. The fact that LookAhead.getScoreAtLevelRec does not work here means that we're doing something wrong there or missing something. Would be good to try to improve LookAhead.getScoreAtLevelRec
This problem does actually not have a perfect solution. This is a heuristics and it will always have something missed.
You can improve it to fix one particular case but there will be eventually another instance of the same problem.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125287/new/

https://reviews.llvm.org/D125287