diff options
| author |  Rasmus Munk Larsen <rmlarsen@google.com> | 2019-05-31 15:26:06 -0700 |
|---|---|---|
| committer | Rasmus Munk Larsen <rmlarsen@google.com> | 2019-05-31 15:26:06 -0700 |
| commit | b08527b0c1ffdbd44347ca3a7869f10b0cb3cbb6 (patch) | |
| tree | 0e96b895f59e4a77ca9880d2e219ffbde11680ac /unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h | |
| parent | 56144005811e3e5a76031ba0aac8a4e1fa3e3396 (diff) | |
Clean up CUDA/NVCC version macros and their use in Eigen, and a few other CUDA build failures.
Diffstat (limited to 'unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h')
| -rw-r--r-- | unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h b/unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h index 7ee4a6087..095bb54cc 100644 --- a/unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h +++ b/unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h @@ -177,7 +177,7 @@ __global__ void FullReductionKernel(Reducer reducer, const Self input, Index num } else { reducer.reduce(__shfl_down(static_cast<int>(accum), offset, warpSize), &accum); } - #elif defined(EIGEN_CUDACC_VER) && EIGEN_CUDACC_VER < 90000 + #elif defined(EIGEN_CUDA_SDK_VER) && EIGEN_CUDA_SDK_VER < 90000 reducer.reduce(__shfl_down(accum, offset, warpSize), &accum); #else reducer.reduce(__shfl_down_sync(0xFFFFFFFF, accum, offset, warpSize), &accum); @@ -269,7 +269,7 @@ __global__ void FullReductionKernelHalfFloat(Reducer reducer, const Self input, wka_in.h = accum; wka_out.i = __shfl_down(wka_in.i, offset, warpSize); reducer.reducePacket(wka_out.h, &accum); - #elif defined(EIGEN_CUDACC_VER) && EIGEN_CUDACC_VER < 90000 + #elif defined(EIGEN_CUDA_SDK_VER) && EIGEN_CUDA_SDK_VER < 90000 reducer.reducePacket(__shfl_down(accum, offset, warpSize), &accum); #else int temp = __shfl_down_sync(0xFFFFFFFF, *(int*)(&accum), (unsigned)offset, warpSize); @@ -466,7 +466,7 @@ __global__ void InnerReductionKernel(Reducer reducer, const Self input, Index nu } else { reducer.reduce(__shfl_down(static_cast<int>(reduced_val), offset), &reduced_val); } - #elif defined(EIGEN_CUDACC_VER) && EIGEN_CUDACC_VER < 90000 + #elif defined(EIGEN_CUDA_SDK_VER) && EIGEN_CUDA_SDK_VER < 90000 reducer.reduce(__shfl_down(reduced_val, offset), &reduced_val); #else reducer.reduce(__shfl_down_sync(0xFFFFFFFF, reduced_val, offset), &reduced_val); @@ -571,7 +571,7 @@ __global__ void InnerReductionKernelHalfFloat(Reducer reducer, const Self input, wka_in.h = reduced_val2; wka_out.i = __shfl_down(wka_in.i, offset, warpSize); reducer.reducePacket(wka_out.h, &reduced_val2); - #elif defined(EIGEN_CUDACC_VER) && EIGEN_CUDACC_VER < 90000 + #elif defined(EIGEN_CUDA_SDK_VER) && EIGEN_CUDA_SDK_VER < 90000 reducer.reducePacket(__shfl_down(reduced_val1, offset, warpSize), &reduced_val1); reducer.reducePacket(__shfl_down(reduced_val2, offset, warpSize), &reduced_val2); #else |