From: Adam Nemet <anemet@apple.com>
Date: Wed, 27 Jan 2016 22:21:25 +0000 (+0000)
Subject: [TTI] Add getPrefetchDistance from PPCLoopDataPrefetch, NFC
X-Git-Tag: llvmorg-3.9.0-rc1~15795
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=dadfbb52f771fc3d3dc9721714598630f5d53146;p=platform%2Fupstream%2Fllvm.git

[TTI] Add getPrefetchDistance from PPCLoopDataPrefetch, NFC

This patch is part of the work to make PPCLoopDataPrefetch
target-independent
(http://thread.gmane.org/gmane.comp.compilers.llvm.devel/92758).

As it was discussed in the above thread, getPrefetchDistance is
currently using instruction count which may change in the future.

llvm-svn: 258995
---

diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index 30bda63..3125853 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -419,6 +419,10 @@ public:
   /// \return The size of a cache line in bytes.
   unsigned getCacheLineSize() const;
 
+  /// \return How much before a load we should place the prefetch instruction.
+  /// This is currently measured in number of instructions.
+  unsigned getPrefetchDistance() const;
+
   /// \return The maximum interleave factor that any transform should try to
   /// perform for this target. This number depends on the level of parallelism
   /// and the number of execution units in the CPU.
@@ -613,6 +617,7 @@ public:
   virtual unsigned getNumberOfRegisters(bool Vector) = 0;
   virtual unsigned getRegisterBitWidth(bool Vector) = 0;
   virtual unsigned getCacheLineSize() = 0;
+  virtual unsigned getPrefetchDistance() = 0;
   virtual unsigned getMaxInterleaveFactor(unsigned VF) = 0;
   virtual unsigned
   getArithmeticInstrCost(unsigned Opcode, Type *Ty, OperandValueKind Opd1Info,
@@ -782,6 +787,7 @@ public:
   unsigned getCacheLineSize() override {
     return Impl.getCacheLineSize();
   }
+  unsigned getPrefetchDistance() override { return Impl.getPrefetchDistance(); }
   unsigned getMaxInterleaveFactor(unsigned VF) override {
     return Impl.getMaxInterleaveFactor(VF);
   }
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
index f4257da..ec797b1 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -266,6 +266,8 @@ public:
 
   unsigned getCacheLineSize() { return 0; }
 
+  unsigned getPrefetchDistance() { return 0; }
+
   unsigned getMaxInterleaveFactor(unsigned VF) { return 1; }
 
   unsigned getArithmeticInstrCost(unsigned Opcode, Type *Ty,
diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp
index ed7005e..bd95789 100644
--- a/llvm/lib/Analysis/TargetTransformInfo.cpp
+++ b/llvm/lib/Analysis/TargetTransformInfo.cpp
@@ -219,6 +219,10 @@ unsigned TargetTransformInfo::getCacheLineSize() const {
   return TTIImpl->getCacheLineSize();
 }
 
+unsigned TargetTransformInfo::getPrefetchDistance() const {
+  return TTIImpl->getPrefetchDistance();
+}
+
 unsigned TargetTransformInfo::getMaxInterleaveFactor(unsigned VF) const {
   return TTIImpl->getMaxInterleaveFactor(VF);
 }
diff --git a/llvm/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp b/llvm/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp
index c113ae5..14553ab 100644
--- a/llvm/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp
+++ b/llvm/lib/Target/PowerPC/PPCLoopDataPrefetch.cpp
@@ -44,12 +44,6 @@ static cl::opt<bool>
 PrefetchWrites("ppc-loop-prefetch-writes", cl::Hidden, cl::init(false),
                cl::desc("Prefetch write addresses"));
 
-// This seems like a reasonable default for the BG/Q (this pass is enabled, by
-// default, only on the BG/Q).
-static cl::opt<unsigned>
-PrefDist("ppc-loop-prefetch-distance", cl::Hidden, cl::init(300),
-         cl::desc("The loop prefetch distance"));
-
 namespace llvm {
   void initializePPCLoopDataPrefetchPass(PassRegistry&);
 }
@@ -107,6 +101,8 @@ bool PPCLoopDataPrefetch::runOnFunction(Function &F) {
   TTI = &getAnalysis<TargetTransformInfoWrapperPass>().getTTI(F);
 
   assert(TTI->getCacheLineSize() && "Cache line size is not set for target");
+  assert(TTI->getPrefetchDistance() &&
+         "Prefetch distance is not set for target");
 
   bool MadeChange = false;
 
@@ -147,7 +143,7 @@ bool PPCLoopDataPrefetch::runOnLoop(Loop *L) {
   if (!LoopSize)
     LoopSize = 1;
 
-  unsigned ItersAhead = PrefDist/LoopSize;
+  unsigned ItersAhead = TTI->getPrefetchDistance() / LoopSize;
   if (!ItersAhead)
     ItersAhead = 1;
 
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index b60c07c..9212e91 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -27,6 +27,12 @@ static cl::opt<unsigned>
 CacheLineSize("ppc-loop-prefetch-cache-line", cl::Hidden, cl::init(64),
               cl::desc("The loop prefetch cache line size"));
 
+// This seems like a reasonable default for the BG/Q (this pass is enabled, by
+// default, only on the BG/Q).
+static cl::opt<unsigned>
+PrefDist("ppc-loop-prefetch-distance", cl::Hidden, cl::init(300),
+         cl::desc("The loop prefetch distance"));
+
 //===----------------------------------------------------------------------===//
 //
 // PPC cost model.
@@ -242,6 +248,8 @@ unsigned PPCTTIImpl::getCacheLineSize() {
   return CacheLineSize;
 }
 
+unsigned PPCTTIImpl::getPrefetchDistance() { return PrefDist; }
+
 unsigned PPCTTIImpl::getMaxInterleaveFactor(unsigned VF) {
   unsigned Directive = ST->getDarwinDirective();
   // The 440 has no SIMD support, but floating-point instructions
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
index d216bdf..5ea9a54 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -71,6 +71,7 @@ public:
   unsigned getNumberOfRegisters(bool Vector);
   unsigned getRegisterBitWidth(bool Vector);
   unsigned getCacheLineSize();
+  unsigned getPrefetchDistance();
   unsigned getMaxInterleaveFactor(unsigned VF);
   int getArithmeticInstrCost(
       unsigned Opcode, Type *Ty,