[PATCH] D98614: [AMDGPU] Fix shortfalls in WQM marking

[Carl Ritson via Phabricator via llvm-commits]

critson marked 6 inline comments as done.
critson added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIWholeQuadMode.cpp:368-370
+      for (; PI != PE && !NextValue; ++PI, ++Idx) {
+        if (Idx < NextPredIdx)
+          continue;
----------------
piotr wrote:
> critson wrote:
> > piotr wrote:
> > > Instead of starting with Idx = 0 and skipping, can you start with Idx = NextPredIdx? That would remove the need for the "continue".
> > What is happening here is we are skipping through the list of predecessor blocks to the correct number.  I do not think there is a way to access a specific predecessor by index?
> auto PI = MBB->pred_begin() + NextPredIdx; ?
Thanks!
I didn't know you could do that with iterators.
(You can see the limits of my C++ knowledge.)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98614/new/

https://reviews.llvm.org/D98614