Displaying 1 result from an estimated 1 matches for "opsmodified".
2016 Aug 03
6
[SCEV] getMulExpr could be extremely slow when creating SCEVs for a long chain of add/mul instructions
...ays integers with values
// known at compile time, never SCEV objects.
//
// The implementation avoids pointless extra computations when the two
// addrec's are of different length (mathematically, it's equivalent to
// an infinite stream of zeros on the right).
bool OpsModified = false;
for (unsigned OtherIdx = Idx+1;
OtherIdx != Ops.size() && isa<SCEVAddRecExpr>(Ops[OtherIdx]);
...
...
One way I thought about to fix this is to have a threshold for the number of terms in each AddRec when folding them. For example, we only fold AddRec1 * AddRec2...