diff options
| author |  Benoit Steiner <benoit.steiner.goog@gmail.com> | 2015-12-17 13:39:01 -0800 |
|---|---|---|
| committer | Benoit Steiner <benoit.steiner.goog@gmail.com> | 2015-12-17 13:39:01 -0800 |
| commit | 4aac55f684d9bd36b5f855fa5a8c2f17ca3094c9 (patch) | |
| tree | 7d70c007a42527cf5e8ddcf6ae26161b7df1d5b5 /unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h | |
| parent | 40e6250fc3737ff76224b04c94c2de3ce0d51607 (diff) | |
Silenced some compilation warnings triggered by nvcc
Diffstat (limited to 'unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h')
| -rw-r--r-- | unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h | 16 |
1 files changed, 12 insertions, 4 deletions
diff --git a/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h b/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h index d93e1de1b..c28078882 100644 --- a/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h +++ b/unsupported/Eigen/CXX11/src/Tensor/TensorExecutor.h @@ -156,14 +156,14 @@ template <typename Expression> class TensorExecutor<Expression, GpuDevice, false> { public: typedef typename Expression::Index Index; - static void run(const Expression& expr, const GpuDevice& device); + EIGEN_DEVICE_FUNC static void run(const Expression& expr, const GpuDevice& device); }; template <typename Expression> class TensorExecutor<Expression, GpuDevice, true> { public: typedef typename Expression::Index Index; - static void run(const Expression& expr, const GpuDevice& device); + EIGEN_DEVICE_FUNC static void run(const Expression& expr, const GpuDevice& device); }; #if defined(__CUDACC__) @@ -213,8 +213,9 @@ EigenMetaKernel_Vectorizable(Evaluator memcopied_eval, Index size) { /*static*/ template <typename Expression> -inline void TensorExecutor<Expression, GpuDevice, false>::run(const Expression& expr, const GpuDevice& device) +EIGEN_DEVICE_FUNC inline void TensorExecutor<Expression, GpuDevice, false>::run(const Expression& expr, const GpuDevice& device) { +#ifndef __CUDA_ARCH__ TensorEvaluator<Expression, GpuDevice> evaluator(expr, device); const bool needs_assign = evaluator.evalSubExprsIfNeeded(NULL); if (needs_assign) @@ -227,13 +228,17 @@ inline void TensorExecutor<Expression, GpuDevice, false>::run(const Expression& LAUNCH_CUDA_KERNEL((EigenMetaKernel_NonVectorizable<TensorEvaluator<Expression, GpuDevice>, Index>), num_blocks, block_size, 0, device, evaluator, size); } evaluator.cleanup(); +#else + eigen_assert(false && "Cannot launch a kernel from another kernel"); +#endif } /*static*/ template<typename Expression> -inline void TensorExecutor<Expression, GpuDevice, true>::run(const Expression& expr, const GpuDevice& device) +EIGEN_DEVICE_FUNC inline void TensorExecutor<Expression, GpuDevice, true>::run(const Expression& expr, const GpuDevice& device) { +#ifndef __CUDA_ARCH__ TensorEvaluator<Expression, GpuDevice> evaluator(expr, device); const bool needs_assign = evaluator.evalSubExprsIfNeeded(NULL); if (needs_assign) @@ -246,6 +251,9 @@ inline void TensorExecutor<Expression, GpuDevice, true>::run(const Expression& e LAUNCH_CUDA_KERNEL((EigenMetaKernel_Vectorizable<TensorEvaluator<Expression, GpuDevice>, Index>), num_blocks, block_size, 0, device, evaluator, size); } evaluator.cleanup(); +#else + eigen_assert(false && "Cannot launch a kernel from another kernel"); +#endif } #endif // __CUDACC__ |