Updates corresponding to the latest round of PR feedback

The major changes are

1. Moving CUDA/PacketMath.h to GPU/PacketMath.h
2. Moving CUDA/MathFunctions.h to GPU/MathFunction.h
3. Moving CUDA/CudaSpecialFunctions.h to GPU/GpuSpecialFunctions.h
    The above three changes effectively enable the Eigen "Packet" layer for the HIP platform

4. Merging the "hip_basic" and "cuda_basic" unit tests into one ("gpu_basic")
5. Updating the "EIGEN_DEVICE_FUNC" marking in some places

The change has been tested on the HIP and CUDA platforms.

1 files changed, 3 insertions, 8 deletions

diff --git a/unsupported/Eigen/CXX11/src/Tensor/TensorReduction.h b/unsupported/Eigen/CXX11/src/Tensor/TensorReduction.h
index fdd338b96..ce573d730 100644
--- a/unsupported/Eigen/CXX11/src/Tensor/TensorReduction.h
+++ b/unsupported/Eigen/CXX11/src/Tensor/TensorReduction.h
@@ -497,6 +497,9 @@ struct TensorEvaluator<const TensorReductionOp<Op, Dims, ArgType, MakePointer_>,
 
   EIGEN_STRONG_INLINE
     #if !defined(EIGEN_HIPCC)
+    // Marking this as EIGEN_DEVICE_FUNC for HIPCC requires also doing the same for all the functions
+    // being called within here, which then leads to proliferation of EIGEN_DEVICE_FUNC markings, one
+    // of which will eventually result in an NVCC error
     EIGEN_DEVICE_FUNC
     #endif
     bool evalSubExprsIfNeeded(typename MakePointer_<CoeffReturnType>::Type data) {
@@ -778,17 +781,9 @@ struct TensorEvaluator<const TensorReductionOp<Op, Dims, ArgType, MakePointer_>,
   // Indexed by reduced dimensions.
   array<Index, NumReducedDims> m_reducedDims;
 
-#if defined(EIGEN_HIPCC)
- public:
-#endif
-  
   // Evaluator for the input expression.
   TensorEvaluator<ArgType, Device> m_impl;
 
-#if defined(EIGEN_HIPCC)
- private:
-#endif
-  
   // Operation to apply for computing the reduction.
   Op m_reducer;