[PATCH] D138748: [VPlan] Summarize recipes used to model inductions.

[Ayal Zaks via Phabricator via llvm-commits]

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

Ship it, thanks!
Add [NFC] to title?
Relate to recent IV recipe patches in commit message?



================
Comment at: llvm/lib/Transforms/Vectorize/VPlan.h:1138
+///    controls exiting of the vector loop by comparing against the vector trip
+///    count. Produces a scalar PHI for the induction value per iteration.
+///  * VPWidenIntOrFpInductionRecipe: Generates vector values for integer and
----------------
nit: worth emphasizing - "a [single] scalar PHI"


================
Comment at: llvm/lib/Transforms/Vectorize/VPlan.h:1141
+///    floating point inductions with arbitrary start and step values. Produces
+///    vector phis for each part.
+///  * VPDerivedIVRecipe: Converts the canonical IV value to the corresponding
----------------
nit: "vector phis for each part" >> "a vector PHI per-part", to be more consistent with above and below?


================
Comment at: llvm/lib/Transforms/Vectorize/VPlan.h:1143
+///  * VPDerivedIVRecipe: Converts the canonical IV value to the corresponding
+///  value of an IV with different start and step values.
+///  * VPScalarIVStepsRecipe: Generates per-lane scalar values based on a
----------------
"Produces a single scalar value per iteration"


================
Comment at: llvm/lib/Transforms/Vectorize/VPlan.h:1147
+///  * VPWidenPointerInductionRecipe: Generate vector and scalar values for a
+///    pointer induction. Produces either vector phis or scalar values based on
+///    the canonical induction.
----------------
"vector phis" - better use either "phi" or "PHI" consistently.

Worth adding that vector phis are produced per-part and scalar values per-lane?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138748/new/

https://reviews.llvm.org/D138748