[PATCH] D43876: [LoopUnroll] Peel off iterations if it makes conditions true/false.
Max Kazantsev via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Mon Mar 12 22:02:00 PDT 2018
mkazantsev added inline comments.
================
Comment at: lib/Transforms/Utils/LoopUnrollPeel.cpp:173
}
+
+ // Pay respect to limitations implied by loop size and the max peel count.
+ unsigned MaxPeelCount = UnrollPeelMaxCount;
+ MaxPeelCount = std::min(MaxPeelCount, UP.Threshold / LoopSize - 1);
+
+ DesiredPeelCount = std::max(DesiredPeelCount,
+ countToEliminateCompares(*L, MaxPeelCount, SE));
+
if (DesiredPeelCount > 0) {
- // Pay respect to limitations implied by loop size and the max peel count.
----------------
fhahn wrote:
> mkazantsev wrote:
> > What was the point of moving this `if`? Could we not just update DesiredPeelCount before this line?
> > We only need MaxPeelCount under this condition, there is no point in calculating it before it.
> MaxPeelCount is passed to `countToEliminateCompares`, to limit the maximum iterations we do.
My bad, I didn't notice it. :)
================
Comment at: lib/Transforms/Utils/LoopUnrollPeel.cpp:194
+ SE.isKnownPredicate(Pred, IterVal, RightSCEV)) {
+ IterVal = SE.getAddExpr(
+ IterVal, cast<SCEVAddRecExpr>(LeftSCEV)->getStepRecurrence(SE));
----------------
fhahn wrote:
> mkazantsev wrote:
> > Step calculation can be hoisted out of this loop. I would also suggest bailing early if AR is not affine because adding a step of non-affine AddRec many times can produce really big and ugly SCEVs.
> I've added a comment that makes it clearer I hope. The idea is to handle cases like below, where the condition is known to be false initially. Initially `i > 2` is not known, but the inverse `i <= 2` is known.
>
> ```
> if (i > 2) {
> // do something
> } else {
> // do something else
> }
> ```
I see the point now, but you are using wrong function. If you look into `getSwappedPredicate`, it returns `<` for `>`, and what you need is called `getInversePredicate`.
https://reviews.llvm.org/D43876
More information about the llvm-commits
mailing list