@Ayal , would be nice if you find time to look at this patch since next two (accepted) patched depends on this one.

Fixed comments.

Rebase

Moved regression test to a separate review.

ping @Ayal

ping @Ayal

Introduce control flag for the feature and disable by default + Rebase.

ebrevnov committed rGf0abe820eebf: [LoopUtils][NFC] Minor refactoring in getLoopEstimatedTripCount. (authored by ebrevnov).

Rebase

Fixed raised issue + Rebase.

Rebase

Comments addressed.

Factor out NFC stuff + rebase

One more round of updates.

Rebase.

ebrevnov committed rG948e745270de: [LV][NFC] Keep dominator tree up to date during vectorization. (authored by ebrevnov).

ebrevnov committed rG1b6286b945a5: [LV][NFC] Some refactoring and renaming to facilitate next change. (authored by ebrevnov).

Hi @Ayal. Thanks for you input. I fixed all places as you suggested. Please check.

Updated as requested.

Rebase

Rebase

Updated according to comments.

In D71047#1793952, @fhahn wrote:

LGTM thanks!

ping @fhahn, @Ayal, @hsaito @anna

ping @Ayal

Rebase.

Tests update

Ping @fhahn

Nope.

Ping

Minor update.

In D71249#1779171, @rengolin wrote:

This change may conflict with your other two. How are we supposed to review them? Are they a set?

If so, what's the order in which they have to be applied?

Yes there are a dependencies between change sets. Especially this and next one. The order is defined by parent/child relationship and can be found at "Revision Contents" section Stack tab.

In D71250#1777947, @rengolin wrote:

This change introduces a number of new branches in the code. While not necessarily the hottest code, it may prove significant on short loops, especially nested loops. Without evidence in the form of benchmarks etc. it's hard to justify.

These extra branches are very short leaved and optimized out by SimplifyCFG which happens just 3 passes after the vectorizer. Thus I don't believe it can cause any harm to the performance. I double check that LLVM's test-suite has no regressions.

Updating as requested by Florian.

In D71047#1773476, @fhahn wrote:

IIUC ideally there should be no changes to the tests required to update the DT handling.

Typo fixed.

Addressing issues raised by Ayal.

In D67905#1770563, @Ayal wrote:

Still think it would be better to provide this as a standalone function in Transforms/Utils/LoopUtils, for potential benefit of loop unroll (and jam) passes in addition to LV. Having agreed to ignore foldTail and requiresScalarEpilog, there's nothing vectorization-specific to do here. There's still an issue though with the fact that LV may use the scalar loop for both the remaining TC%(VF*UF) iterations when running the vector loop, and for all TC iterations when runtime guards bypass the vector loop. In absence of information, each such guard could be assigned 0.5 probability, or one could be aggressively optimistic and hope vector loop is always reached. In any case this deserves a comment.

Suggesting further variable name changes for the three Orig, Unrolled, and Remainder Loops, each having a LoopEntry==LoopExit edge weight, a Backedge weight, a HeaderBlock weight, and an AverageTripCount. The actual weights are recorded as TrueVal and FalseVal of the latch branches.

Patch needs to be clang-format'ed.

Minor naming update

Use SplitBlock utility to automatically udpate DT and LI.

Minor formatting update.

In D71053#1770322, @rengolin wrote:

Is this change inspired by a real world case? If so, how relevant / pervasive is this case?

Yes, this case came from the real world benchmark. This change gives +15% on it.

Restore lost change.

Extracted formatting and renaming to a separate review.