[PATCH] D103788: [InstCombine] Eliminate casts to optimize ctlz operation

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri requested changes to this revision.
lebedev.ri added a comment.
This revision now requires changes to proceed.

I think this is being approached from the wrong angle.
You currently transform `trunc (ctlz(zext(A))) --> add(ctlz(A), (bitwidth(zext(A))-bitwidth(A))`,
but why does that `trunc` matter? 
Wouldn't it make more sense to view this as `ctlz(zext(A)) --> add(ctlz(A), (bitwidth(zext(A))-bitwidth(A))`?



================
Comment at: llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp:954
+  // trunc (ctlz_i32(zext(A), B) --> add(ctlz_i16(A, B), C)
+  if (match(Src, m_OneUse(m_Intrinsic<Intrinsic::ctlz>(
+                     m_OneUse(m_ZExt(m_Value(A))), m_Value(B))))) {
----------------
Are you not missing a check that types of `A` and `Trunc` match?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103788/new/

https://reviews.llvm.org/D103788