[NVPTX] Added support for half-precision floating point.

Only scalar half-precision operations are supported at the moment.

- Adds general support for 'half' type in NVPTX.
- fp16 math operations are supported on sm_53+ GPUs only
  (can be disabled with --nvptx-no-f16-math).
- Type conversions to/from fp16 are supported on all GPU variants.
- On GPU variants that do not have full fp16 support (or if it's disabled),
  fp16 operations are promoted to fp32 and results are converted back
  to fp16 for storage.

Differential Revision: https://reviews.llvm.org/D28540

llvm-svn: 291956

1 files changed, 7 insertions, 0 deletions

diff --git a/llvm/lib/Target/NVPTX/NVPTXMCExpr.cpp b/llvm/lib/Target/NVPTX/NVPTXMCExpr.cpp
index eab5ee80561..86a28f7d070 100644
--- a/llvm/lib/Target/NVPTX/NVPTXMCExpr.cpp
+++ b/llvm/lib/Target/NVPTX/NVPTXMCExpr.cpp
@@ -27,6 +27,13 @@ void NVPTXFloatMCExpr::printImpl(raw_ostream &OS, const MCAsmInfo *MAI) const {
 
   switch (Kind) {
   default: llvm_unreachable("Invalid kind!");
+  case VK_NVPTX_HALF_PREC_FLOAT:
+    // ptxas does not have a way to specify half-precision floats.
+    // Instead we have to print and load fp16 constants as .b16
+    OS << "0x";
+    NumHex = 4;
+    APF.convert(APFloat::IEEEhalf(), APFloat::rmNearestTiesToEven, &Ignored);
+    break;
   case VK_NVPTX_SINGLE_PREC_FLOAT:
     OS << "0f";
     NumHex = 8;