[flang] Add PPC vec_max, vec_min, vec_madd and vec_nmsub intrinsic

Differential Revision: https://reviews.llvm.org/D152938

Author: kkwli

flang/include/flang/Optimizer/Builder/IntrinsicCall.h

@@ -481,8 +481,19 @@ enum class ParamTypeId {
   Integer,
   Real,
   Complex,
+  IntegerVector,
+  UnsignedVector,
+  RealVector,
 };
 
+// Helper function to get length of a 16-byte vector of element type eleTy.
+static int getVecLen(mlir::Type eleTy) {
+  assert((mlir::isa<mlir::IntegerType>(eleTy) ||
+          mlir::isa<mlir::FloatType>(eleTy)) &&
+         "unsupported vector element type");
+  return 16 / (eleTy.getIntOrFloatBitWidth() / 8);
+}
+
 template <ParamTypeId t, int k>
 struct ParamType {
   // Supported kinds can be checked with static asserts at compile time.
@@ -509,6 +520,12 @@ template <int k>
 using Integer = ParamType<ParamTypeId::Integer, k>;
 template <int k>
 using Complex = ParamType<ParamTypeId::Complex, k>;
+template <int k>
+using IntegerVector = ParamType<ParamTypeId::IntegerVector, k>;
+template <int k>
+using RealVector = ParamType<ParamTypeId::RealVector, k>;
+template <int k>
+using UnsignedVector = ParamType<ParamTypeId::UnsignedVector, k>;
 } // namespace Ty
 
 // Helper function that generates most types that are supported for intrinsic
@@ -518,24 +535,46 @@ static inline mlir::Type getTypeHelper(mlir::MLIRContext *context,
                                        fir::FirOpBuilder &builder,
                                        ParamTypeId typeId, int kind) {
   mlir::Type r;
-  int bits = 0;
+  unsigned bits{0};
   switch (typeId) {
   case ParamTypeId::Void:
     llvm::report_fatal_error("can not get type of void");
     break;
   case ParamTypeId::Integer:
+  case ParamTypeId::IntegerVector:
     bits = builder.getKindMap().getIntegerBitsize(kind);
     assert(bits != 0 && "failed to convert kind to integer bitsize");
     r = mlir::IntegerType::get(context, bits);
     break;
+  case ParamTypeId::UnsignedVector:
+    bits = builder.getKindMap().getIntegerBitsize(kind);
+    assert(bits != 0 && "failed to convert kind to unsigned bitsize");
+    r = mlir::IntegerType::get(context, bits, mlir::IntegerType::Unsigned);
+    break;
   case ParamTypeId::Real:
+  case ParamTypeId::RealVector:
     r = builder.getRealType(kind);
     break;
   case ParamTypeId::Complex:
     r = fir::ComplexType::get(context, kind);
     break;
   }
-  return r;
+
+  switch (typeId) {
+  case ParamTypeId::Void:
+  case ParamTypeId::Integer:
+  case ParamTypeId::Real:
+  case ParamTypeId::Complex:
+    // keep original type for void and non-vector
+    return r;
+    break;
+  case ParamTypeId::IntegerVector:
+  case ParamTypeId::UnsignedVector:
+  case ParamTypeId::RealVector:
+    // convert to FIR vector type
+    return fir::VectorType::get(getVecLen(r), r);
+    break;
+  }
 }
 
 // Generic function type generator that supports most of the function types

flang/lib/Optimizer/Builder/PPCIntrinsicCall.cpp

@@ -116,6 +116,98 @@ static constexpr MathOperation ppcMathOperations[] = {
      genLibCall},
     {"__ppc_frsqrtes", "llvm.ppc.frsqrtes",
      genFuncType<Ty::Real<4>, Ty::Real<4>>, genLibCall},
+    {"__ppc_vec_madd", "llvm.fma.v4f32",
+     genFuncType<Ty::RealVector<4>, Ty::RealVector<4>, Ty::RealVector<4>,
+                 Ty::RealVector<4>>,
+     genLibCall},
+    {"__ppc_vec_madd", "llvm.fma.v2f64",
+     genFuncType<Ty::RealVector<8>, Ty::RealVector<8>, Ty::RealVector<8>,
+                 Ty::RealVector<8>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxsb",
+     genFuncType<Ty::IntegerVector<1>, Ty::IntegerVector<1>,
+                 Ty::IntegerVector<1>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxsh",
+     genFuncType<Ty::IntegerVector<2>, Ty::IntegerVector<2>,
+                 Ty::IntegerVector<2>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxsw",
+     genFuncType<Ty::IntegerVector<4>, Ty::IntegerVector<4>,
+                 Ty::IntegerVector<4>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxsd",
+     genFuncType<Ty::IntegerVector<8>, Ty::IntegerVector<8>,
+                 Ty::IntegerVector<8>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxub",
+     genFuncType<Ty::UnsignedVector<1>, Ty::UnsignedVector<1>,
+                 Ty::UnsignedVector<1>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxuh",
+     genFuncType<Ty::UnsignedVector<2>, Ty::UnsignedVector<2>,
+                 Ty::UnsignedVector<2>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxuw",
+     genFuncType<Ty::UnsignedVector<4>, Ty::UnsignedVector<4>,
+                 Ty::UnsignedVector<4>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.altivec.vmaxud",
+     genFuncType<Ty::UnsignedVector<8>, Ty::UnsignedVector<8>,
+                 Ty::UnsignedVector<8>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.vsx.xvmaxsp",
+     genFuncType<Ty::RealVector<4>, Ty::RealVector<4>, Ty::RealVector<4>>,
+     genLibCall},
+    {"__ppc_vec_max", "llvm.ppc.vsx.xvmaxdp",
+     genFuncType<Ty::RealVector<8>, Ty::RealVector<8>, Ty::RealVector<8>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminsb",
+     genFuncType<Ty::IntegerVector<1>, Ty::IntegerVector<1>,
+                 Ty::IntegerVector<1>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminsh",
+     genFuncType<Ty::IntegerVector<2>, Ty::IntegerVector<2>,
+                 Ty::IntegerVector<2>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminsw",
+     genFuncType<Ty::IntegerVector<4>, Ty::IntegerVector<4>,
+                 Ty::IntegerVector<4>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminsd",
+     genFuncType<Ty::IntegerVector<8>, Ty::IntegerVector<8>,
+                 Ty::IntegerVector<8>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminub",
+     genFuncType<Ty::UnsignedVector<1>, Ty::UnsignedVector<1>,
+                 Ty::UnsignedVector<1>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminuh",
+     genFuncType<Ty::UnsignedVector<2>, Ty::UnsignedVector<2>,
+                 Ty::UnsignedVector<2>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminuw",
+     genFuncType<Ty::UnsignedVector<4>, Ty::UnsignedVector<4>,
+                 Ty::UnsignedVector<4>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.altivec.vminud",
+     genFuncType<Ty::UnsignedVector<8>, Ty::UnsignedVector<8>,
+                 Ty::UnsignedVector<8>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.vsx.xvminsp",
+     genFuncType<Ty::RealVector<4>, Ty::RealVector<4>, Ty::RealVector<4>>,
+     genLibCall},
+    {"__ppc_vec_min", "llvm.ppc.vsx.xvmindp",
+     genFuncType<Ty::RealVector<8>, Ty::RealVector<8>, Ty::RealVector<8>>,
+     genLibCall},
+    {"__ppc_vec_nmsub", "llvm.ppc.fnmsub.v4f32",
+     genFuncType<Ty::RealVector<4>, Ty::RealVector<4>, Ty::RealVector<4>,
+                 Ty::RealVector<4>>,
+     genLibCall},
+    {"__ppc_vec_nmsub", "llvm.ppc.fnmsub.v2f64",
+     genFuncType<Ty::RealVector<8>, Ty::RealVector<8>, Ty::RealVector<8>,
+                 Ty::RealVector<8>>,
+     genLibCall},
 };
 
 const IntrinsicHandler *findPPCIntrinsicHandler(llvm::StringRef name) {

flang/module/__ppc_intrinsics.f90

@@ -49,6 +49,17 @@ elemental vector(real(VKIND)) function elem_func_vr##VKIND##vr##VKIND##vr##VKIND
 #undef ELEM_FUNC_VUVUVU
 #undef ELEM_FUNC_VIVIVI
 
+!! ================ 3 arguments function interface ================
+! vector(r) function f(vector(r), vector(r), vector(r))
+#define ELEM_FUNC_VRVRVRVR(VKIND) \
+  elemental vector(real(VKIND)) function elem_func_vr##VKIND##vr##VKIND##vr##VKIND##vr##VKIND(arg1, arg2, arg3); \
+    vector(real(VKIND)), intent(in) :: arg1, arg2, arg3; \
+  end function ;
+
+  ELEM_FUNC_VRVRVRVR(4) ELEM_FUNC_VRVRVRVR(8)
+
+#undef ELEM_FUNC_VRVRVRVR
+
   end interface
 
   procedure(func_r4r4r4r4) :: __ppc_fmadd_r4
@@ -242,6 +253,28 @@ end function func_r8r8i
   end interface vec_and
   public :: vec_and
 
+! vec_max
+  VEC_VI_VI_VI(vec_max,1) VEC_VI_VI_VI(vec_max,2) VEC_VI_VI_VI(vec_max,4) VEC_VI_VI_VI(vec_max,8)
+  VEC_VU_VU_VU(vec_max,1) VEC_VU_VU_VU(vec_max,2) VEC_VU_VU_VU(vec_max,4) VEC_VU_VU_VU(vec_max,8)
+  VEC_VR_VR_VR(vec_max,4) VEC_VR_VR_VR(vec_max,8)
+  interface vec_max
+    procedure :: VI_VI_VI(vec_max,1), VI_VI_VI(vec_max,2), VI_VI_VI(vec_max,4), VI_VI_VI(vec_max,8)
+    procedure :: VU_VU_VU(vec_max,1), VU_VU_VU(vec_max,2), VU_VU_VU(vec_max,4), VU_VU_VU(vec_max,8)
+    procedure :: VR_VR_VR(vec_max,4), VR_VR_VR(vec_max,8)
+  end interface vec_max
+  public :: vec_max
+
+! vec_min
+  VEC_VI_VI_VI(vec_min,1) VEC_VI_VI_VI(vec_min,2) VEC_VI_VI_VI(vec_min,4) VEC_VI_VI_VI(vec_min,8)
+  VEC_VU_VU_VU(vec_min,1) VEC_VU_VU_VU(vec_min,2) VEC_VU_VU_VU(vec_min,4) VEC_VU_VU_VU(vec_min,8)
+  VEC_VR_VR_VR(vec_min,4) VEC_VR_VR_VR(vec_min,8)
+  interface vec_min
+    procedure :: VI_VI_VI(vec_min,1), VI_VI_VI(vec_min,2), VI_VI_VI(vec_min,4), VI_VI_VI(vec_min,8)
+    procedure :: VU_VU_VU(vec_min,1), VU_VU_VU(vec_min,2), VU_VU_VU(vec_min,4), VU_VU_VU(vec_min,8)
+    procedure :: VR_VR_VR(vec_min,4), VR_VR_VR(vec_min,8)
+  end interface vec_min
+  public :: vec_min
+
 ! vec_mul
   VEC_VI_VI_VI(vec_mul,1) VEC_VI_VI_VI(vec_mul,2) VEC_VI_VI_VI(vec_mul,4) VEC_VI_VI_VI(vec_mul,8)
   VEC_VU_VU_VU(vec_mul,1) VEC_VU_VU_VU(vec_mul,2) VEC_VU_VU_VU(vec_mul,4) VEC_VU_VU_VU(vec_mul,8)
@@ -281,4 +314,30 @@ end function func_r8r8i
 #undef VR_VR_VR
 #undef VU_VU_VU
 #undef VI_VI_VI
+
+!-----------------------------------------
+! vector function(vector, vector, vector)
+!-----------------------------------------
+#define VR_VR_VR_VR(NAME, VKIND) __ppc_##NAME##_vr##VKIND##vr##VKIND##vr##VKIND##vr##VKIND
+
+#define VEC_VR_VR_VR_VR(NAME, VKIND) \
+  procedure(elem_func_vr##VKIND##vr##VKIND##vr##VKIND##vr##VKIND) :: VR_VR_VR_VR(NAME, VKIND);
+
+! vec_madd
+  VEC_VR_VR_VR_VR(vec_madd,4) VEC_VR_VR_VR_VR(vec_madd,8)
+  interface vec_madd
+    procedure :: VR_VR_VR_VR(vec_madd,4), VR_VR_VR_VR(vec_madd,8)
+  end interface vec_madd
+  public :: vec_madd
+
+! vec_nmsub
+  VEC_VR_VR_VR_VR(vec_nmsub,4) VEC_VR_VR_VR_VR(vec_nmsub,8)
+  interface vec_nmsub
+    procedure :: VR_VR_VR_VR(vec_nmsub,4), VR_VR_VR_VR(vec_nmsub,8)
+  end interface vec_nmsub
+  public :: vec_nmsub
+
+#undef VEC_VR_VR_VR_VR
+#undef VR_VR_VR_VR
+
 end module __ppc_intrinsics