X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTransforms%2FUtils%2FSimplifyLibCalls.cpp;h=df75ed96893dbf5665fb88e9d4f36e32bdee0dce;hb=cc87069c319aed2f95394a76dccfcd6360c08b80;hp=6d3dfd6750a52e5a5f7c60a135317cc01d42a031;hpb=46cdbe20488d4ea26320c96868f5c24f8ba42b0c;p=oota-llvm.git

diff --git a/lib/Transforms/Utils/SimplifyLibCalls.cpp b/lib/Transforms/Utils/SimplifyLibCalls.cpp
index 6d3dfd6750a..df75ed96893 100644
--- a/lib/Transforms/Utils/SimplifyLibCalls.cpp
+++ b/lib/Transforms/Utils/SimplifyLibCalls.cpp
@@ -1058,6 +1058,31 @@ Value *LibCallSimplifier::optimizeCos(CallInst *CI, IRBuilder<> &B) {
   return Ret;
 }
 
+static Value *getPow(Value *InnerChain[33], unsigned Exp, IRBuilder<> &B) {
+  // Multiplications calculated using Addition Chains.
+  // Refer: http://wwwhomes.uni-bielefeld.de/achim/addition_chain.html
+
+  assert(Exp != 0 && "Incorrect exponent 0 not handled");
+
+  if (InnerChain[Exp])
+    return InnerChain[Exp];
+
+  static const unsigned AddChain[33][2] = {
+      {0, 0}, // Unused.
+      {0, 0}, // Unused (base case = pow1).
+      {1, 1}, // Unused (pre-computed).
+      {1, 2},  {2, 2},   {2, 3},  {3, 3},   {2, 5},  {4, 4},
+      {1, 8},  {5, 5},   {1, 10}, {6, 6},   {4, 9},  {7, 7},
+      {3, 12}, {8, 8},   {8, 9},  {2, 16},  {1, 18}, {10, 10},
+      {6, 15}, {11, 11}, {3, 20}, {12, 12}, {8, 17}, {13, 13},
+      {3, 24}, {14, 14}, {4, 25}, {15, 15}, {3, 28}, {16, 16},
+  };
+
+  InnerChain[Exp] = B.CreateFMul(getPow(InnerChain, AddChain[Exp][0], B),
+                                 getPow(InnerChain, AddChain[Exp][1], B));
+  return InnerChain[Exp];
+}
+
 Value *LibCallSimplifier::optimizePow(CallInst *CI, IRBuilder<> &B) {
   Function *Callee = CI->getCalledFunction();
   Value *Ret = nullptr;
@@ -1156,6 +1181,32 @@ Value *LibCallSimplifier::optimizePow(CallInst *CI, IRBuilder<> &B) {
     return B.CreateFMul(Op1, Op1, "pow2");
   if (Op2C->isExactlyValue(-1.0)) // pow(x, -1.0) -> 1.0/x
     return B.CreateFDiv(ConstantFP::get(CI->getType(), 1.0), Op1, "powrecip");
+
+  // In -ffast-math, generate repeated fmul instead of generating pow(x, n).
+  if (unsafeFPMath) {
+    APFloat V = abs(Op2C->getValueAPF());
+    // We limit to a max of 7 fmul(s). Thus max exponent is 32.
+    // This transformation applies to integer exponents only.
+    if (V.compare(APFloat(V.getSemantics(), 32.0)) == APFloat::cmpGreaterThan ||
+        !V.isInteger())
+      return nullptr;
+
+    // We will memoize intermediate products of the Addition Chain.
+    Value *InnerChain[33] = {nullptr};
+    InnerChain[1] = Op1;
+    InnerChain[2] = B.CreateFMul(Op1, Op1);
+
+    // We cannot readily convert a non-double type (like float) to a double.
+    // So we first convert V to something which could be converted to double.
+    bool ignored;
+    V.convert(APFloat::IEEEdouble, APFloat::rmTowardZero, &ignored);
+    Value *FMul = getPow(InnerChain, V.convertToDouble(), B);
+    // For negative exponents simply compute the reciprocal.
+    if (Op2C->isNegative())
+      FMul = B.CreateFDiv(ConstantFP::get(CI->getType(), 1.0), FMul);
+    return FMul;
+  }
+
   return nullptr;
 }
 
@@ -1284,6 +1335,56 @@ Value *LibCallSimplifier::optimizeFMinFMax(CallInst *CI, IRBuilder<> &B) {
   return B.CreateSelect(Cmp, Op0, Op1);
 }
 
+Value *LibCallSimplifier::optimizeLog(CallInst *CI, IRBuilder<> &B) {
+  Function *Callee = CI->getCalledFunction();
+  Value *Ret = nullptr;
+  StringRef Name = Callee->getName();
+  if (UnsafeFPShrink && hasFloatVersion(Name))
+    Ret = optimizeUnaryDoubleFP(CI, B, true);
+  FunctionType *FT = Callee->getFunctionType();
+
+  // Just make sure this has 1 argument of FP type, which matches the
+  // result type.
+  if (FT->getNumParams() != 1 || FT->getReturnType() != FT->getParamType(0) ||
+      !FT->getParamType(0)->isFloatingPointTy())
+    return Ret;
+
+  if (!canUseUnsafeFPMath(CI->getParent()->getParent()))
+    return Ret;
+  Value *Op1 = CI->getArgOperand(0);
+  auto *OpC = dyn_cast<CallInst>(Op1);
+  if (!OpC)
+    return Ret;
+
+  // log(pow(x,y)) -> y*log(x)
+  // This is only applicable to log, log2, log10.
+  if (Name != "log" && Name != "log2" && Name != "log10")
+    return Ret;
+
+  IRBuilder<>::FastMathFlagGuard Guard(B);
+  FastMathFlags FMF;
+  FMF.setUnsafeAlgebra();
+  B.SetFastMathFlags(FMF);
+
+  LibFunc::Func Func;
+  Function *F = OpC->getCalledFunction();
+  if (F && ((TLI->getLibFunc(F->getName(), Func) && TLI->has(Func) &&
+      Func == LibFunc::pow) || F->getIntrinsicID() == Intrinsic::pow))
+    return B.CreateFMul(OpC->getArgOperand(1),
+      EmitUnaryFloatFnCall(OpC->getOperand(0), Callee->getName(), B,
+                           Callee->getAttributes()), "mul");
+
+  // log(exp2(y)) -> y*log(2)
+  if (F && Name == "log" && TLI->getLibFunc(F->getName(), Func) &&
+      TLI->has(Func) && Func == LibFunc::exp2)
+    return B.CreateFMul(
+        OpC->getArgOperand(0),
+        EmitUnaryFloatFnCall(ConstantFP::get(CI->getType(), 2.0),
+                             Callee->getName(), B, Callee->getAttributes()),
+        "logmul");
+  return Ret;
+}
+
 Value *LibCallSimplifier::optimizeSqrt(CallInst *CI, IRBuilder<> &B) {
   Function *Callee = CI->getCalledFunction();
   
@@ -2088,6 +2189,8 @@ Value *LibCallSimplifier::optimizeCall(CallInst *CI) {
       return optimizeExp2(CI, Builder);
     case Intrinsic::fabs:
       return optimizeFabs(CI, Builder);
+    case Intrinsic::log:
+      return optimizeLog(CI, Builder);
     case Intrinsic::sqrt:
       return optimizeSqrt(CI, Builder);
     default:
@@ -2170,6 +2273,12 @@ Value *LibCallSimplifier::optimizeCall(CallInst *CI) {
       return optimizeFWrite(CI, Builder);
     case LibFunc::fputs:
       return optimizeFPuts(CI, Builder);
+    case LibFunc::log:
+    case LibFunc::log10:
+    case LibFunc::log1p:
+    case LibFunc::log2:
+    case LibFunc::logb:
+      return optimizeLog(CI, Builder);
     case LibFunc::puts:
       return optimizePuts(CI, Builder);
     case LibFunc::tan:
@@ -2203,11 +2312,6 @@ Value *LibCallSimplifier::optimizeCall(CallInst *CI) {
     case LibFunc::exp:
     case LibFunc::exp10:
     case LibFunc::expm1:
-    case LibFunc::log:
-    case LibFunc::log10:
-    case LibFunc::log1p:
-    case LibFunc::log2:
-    case LibFunc::logb:
     case LibFunc::sin:
     case LibFunc::sinh:
     case LibFunc::tanh:
@@ -2256,12 +2360,9 @@ void LibCallSimplifier::replaceAllUsesWith(Instruction *I, Value *With) {
 //
 // log, logf, logl:
 //   * log(exp(x))   -> x
-//   * log(x**y)     -> y*log(x)
 //   * log(exp(y))   -> y*log(e)
-//   * log(exp2(y))  -> y*log(2)
 //   * log(exp10(y)) -> y*log(10)
 //   * log(sqrt(x))  -> 0.5*log(x)
-//   * log(pow(x,y)) -> y*log(x)
 //
 // lround, lroundf, lroundl:
 //   * lround(cnst) -> cnst'