From 14925e6b885f8bd8cf448627386d412831f4bf1b Mon Sep 17 00:00:00 2001
From: Nadav Rotem <nrotem@apple.com>
Date: Wed, 9 Jan 2013 22:29:00 +0000
Subject: ARM Cost model: Use the size of vector registers and widest
 vectorizable instruction to determine the max vectorization factor.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172010 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Transforms/Vectorize/LoopVectorize.cpp | 51 ++++++++++++++++++++++++++++--
 1 file changed, 48 insertions(+), 3 deletions(-)

(limited to 'lib/Transforms/Vectorize/LoopVectorize.cpp')

diff --git a/lib/Transforms/Vectorize/LoopVectorize.cpp b/lib/Transforms/Vectorize/LoopVectorize.cpp
index c29f416be7..cde4bb889e 100644
--- a/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -113,9 +113,6 @@ static const unsigned MaxLoopSizeThreshold = 32;
 /// number of pointers. Notice that the check is quadratic!
 static const unsigned RuntimeMemoryCheckThreshold = 4;
 
-/// This is the highest vector width that we try to generate.
-static const unsigned MaxVectorSize = 8;
-
 namespace {
 
 // Forward declarations.
@@ -523,6 +520,10 @@ public:
   /// possible.
   unsigned selectVectorizationFactor(bool OptForSize, unsigned UserVF);
 
+  /// \returns The size (in bits) of the widest type in the code that
+  /// needs to be vectorized. We ignore values that remain scalar such as
+  /// 64 bit loop indices.
+  unsigned getWidestType();
 
   /// \return The most profitable unroll factor.
   /// If UserUF is non-zero then this method finds the best unroll-factor
@@ -2621,6 +2622,20 @@ LoopVectorizationCostModel::selectVectorizationFactor(bool OptForSize,
   unsigned TC = SE->getSmallConstantTripCount(TheLoop, TheLoop->getLoopLatch());
   DEBUG(dbgs() << "LV: Found trip count:"<<TC<<"\n");
 
+  unsigned WidestType = getWidestType();
+  unsigned WidestRegister = TTI.getRegisterBitWidth(true);
+  unsigned MaxVectorSize = WidestRegister / WidestType;
+  DEBUG(dbgs() << "LV: The Widest type: " << WidestType << " bits.\n");
+  DEBUG(dbgs() << "LV: The Widest register is:" << WidestRegister << "bits.\n");
+
+  if (MaxVectorSize == 0) {
+    DEBUG(dbgs() << "LV: The target has no vector registers.\n");
+    return 1;
+  }
+
+  assert(MaxVectorSize <= 32 && "Did not expect to pack so many elements"
+         " into one vector.");
+  
   unsigned VF = MaxVectorSize;
 
   // If we optimize the program for size, avoid creating the tail loop.
@@ -2672,6 +2687,36 @@ LoopVectorizationCostModel::selectVectorizationFactor(bool OptForSize,
   return Width;
 }
 
+unsigned LoopVectorizationCostModel::getWidestType() {
+  unsigned MaxWidth = 8;
+
+  // For each block.
+  for (Loop::block_iterator bb = TheLoop->block_begin(),
+       be = TheLoop->block_end(); bb != be; ++bb) {
+    BasicBlock *BB = *bb;
+
+    // For each instruction in the loop.
+    for (BasicBlock::iterator it = BB->begin(), e = BB->end(); it != e; ++it) {
+      if (Legal->isUniformAfterVectorization(it))
+        continue;
+
+      Type *T = it->getType();
+
+      if (StoreInst *ST = dyn_cast<StoreInst>(it))
+        T = ST->getValueOperand()->getType();
+
+      // PHINodes and pointers are difficult to analyze, but we catch all other
+      // uses of the types in other instructions.
+      if (isa<PHINode>(it) || T->isPointerTy() || T->isVoidTy())
+        continue;
+
+      MaxWidth = std::max(MaxWidth, T->getScalarSizeInBits());
+    }
+  }
+
+  return MaxWidth;
+}
+
 unsigned
 LoopVectorizationCostModel::selectUnrollFactor(bool OptForSize,
                                                unsigned UserUF) {
-- 
cgit v1.2.3-70-g09d2