[X86] Enable intrinsics that convert float and bf16 data to each other

Scalar version :
_mm_cvtsbh_ss , _mm_cvtness_sbh

Vector version:
_mm512_cvtpbh_ps , _mm256_cvtpbh_ps
_mm512_maskz_cvtpbh_ps , _mm256_maskz_cvtpbh_ps
_mm512_mask_cvtpbh_ps , _mm256_mask_cvtpbh_ps

Patch by Shengchen Kan (skan)

Differential Revision: https://reviews.llvm.org/D62363

llvm-svn: 363018

1 files changed, 14 insertions, 0 deletions

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index c6d3aff7bc0..81f3e5664aa 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -9795,6 +9795,18 @@ Value *CodeGenFunction::EmitX86CpuIs(const CallExpr *E) {
   return EmitX86CpuIs(CPUStr);
 }
 
+// Convert a BF16 to a float.
+static Value *EmitX86CvtBF16ToFloatExpr(CodeGenFunction &CGF,
+                                        const CallExpr *E,
+                                        ArrayRef<Value *> Ops) {
+  llvm::Type *Int32Ty = CGF.Builder.getInt32Ty();
+  Value *ZeroExt = CGF.Builder.CreateZExt(Ops[0], Int32Ty);
+  Value *Shl = CGF.Builder.CreateShl(ZeroExt, 16);
+  llvm::Type *ResultType = CGF.ConvertType(E->getType());
+  Value *BitCast = CGF.Builder.CreateBitCast(Shl, ResultType);
+  return BitCast;
+}
+
 Value *CodeGenFunction::EmitX86CpuIs(StringRef CPUStr) {
 
   llvm::Type *Int32Ty = Builder.getInt32Ty();
@@ -11941,6 +11953,8 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
     Intrinsic::ID IID = Intrinsic::x86_avx512bf16_mask_cvtneps2bf16_128;
     return Builder.CreateCall(CGM.getIntrinsic(IID), Ops);
   }
+  case X86::BI__builtin_ia32_cvtsbf162ss_32:
+    return EmitX86CvtBF16ToFloatExpr(*this, E, Ops);
 
   case X86::BI__builtin_ia32_cvtneps2bf16_256_mask:
   case X86::BI__builtin_ia32_cvtneps2bf16_512_mask: {