diff options
| author |  Benoit Steiner <benoit.steiner.goog@gmail.com> | 2016-02-19 06:16:07 +0000 |
|---|---|---|
| committer | Benoit Steiner <benoit.steiner.goog@gmail.com> | 2016-02-19 06:16:07 +0000 |
| commit | 17b9fbed34cefe08b4f63dbe0734e12311eb8669 (patch) | |
| tree | 67258f2de18a1dd273b148afba0cc175c4f011e5 /Eigen/Core | |
| parent | 8ce46f9d8959236c0dfb6dd7dca7423d825f0c59 (diff) | |
Added preliminary support for half floats on CUDA GPU. For now we can simply convert floats into half floats and vice versa
Diffstat (limited to 'Eigen/Core')
| -rw-r--r-- | Eigen/Core | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/Eigen/Core b/Eigen/Core index 63602f4c3..17f864084 100644 --- a/Eigen/Core +++ b/Eigen/Core @@ -200,6 +200,7 @@ #if defined __CUDACC__ #define EIGEN_VECTORIZE_CUDA #include <vector_types.h> + #include <cuda_fp16.h> #endif #if (defined _OPENMP) && (!defined EIGEN_DONT_PARALLELIZE) @@ -329,7 +330,9 @@ using std::ptrdiff_t; #if defined EIGEN_VECTORIZE_CUDA #include "src/Core/arch/CUDA/PacketMath.h" + #include "src/Core/arch/CUDA/PacketMathHalf.h" #include "src/Core/arch/CUDA/MathFunctions.h" + #include "src/Core/arch/CUDA/TypeCasting.h" #endif #include "src/Core/arch/Default/Settings.h" |