[VectorUtils] Query number of sign bits to allow more truncations
When deciding if a vector calculation can be done in a smaller bitwidth, use sign bit information from ValueTracking to add more information and allow more truncations.
llvm-svn: 268921
diff --git a/llvm/lib/Analysis/VectorUtils.cpp b/llvm/lib/Analysis/VectorUtils.cpp
index 23a0de8..2c03f1a 100644
--- a/llvm/lib/Analysis/VectorUtils.cpp
+++ b/llvm/lib/Analysis/VectorUtils.cpp
@@ -320,6 +320,9 @@
SmallPtrSet<Instruction *, 4> InstructionSet;
MapVector<Instruction *, uint64_t> MinBWs;
+ assert(Blocks.size() > 0 && "Must have at least one block!");
+ const DataLayout &DL = Blocks[0]->getModule()->getDataLayout();
+
// Determine the roots. We work bottom-up, from truncs or icmps.
bool SeenExtFromIllegalType = false;
for (auto *BB : Blocks)
@@ -363,12 +366,19 @@
// If we encounter a type that is larger than 64 bits, we can't represent
// it so bail out.
- if (DB.getDemandedBits(I).getBitWidth() > 64)
+ APInt NeededBits = DB.getDemandedBits(I);
+ unsigned BW = NeededBits.getBitWidth();
+ if (BW > 64)
return MapVector<Instruction *, uint64_t>();
- uint64_t V = DB.getDemandedBits(I).getZExtValue();
- DBits[Leader] |= V;
- DBits[I] = V;
+ auto NSB = ComputeNumSignBits(I, DL);
+
+ // Query demanded bits for the bits required by the instruction. Remove
+ // any bits that are equal to the sign bit, because we can truncate the
+ // instruction without changing their value.
+ NeededBits &= APInt::getLowBitsSet(BW, BW - NSB);
+ DBits[Leader] |= NeededBits.getZExtValue();
+ DBits[I] |= NeededBits.getZExtValue();
// Casts, loads and instructions outside of our range terminate a chain
// successfully.