Compile tensorflow and error : initializer not allowed for __shared__ variable
Compile tensorflow and error : initializer not allowed for __shared__ variable
我尝试在我的电脑中编译 tensorflow-1.7.0 并包含软件
cuda-10.0
cudnn-7.3
python-3.6.2,
visual studio 2017
cmake.
有错误：
不知道是怎么发生的，也不知道怎么解决
：“C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_python_build_pip_package.vcxproj”(默认目标) (1) ->
“C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\pywrap_tensorflow_internal.vcxproj”(默认目标) (2) ->
“C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels.vcxproj”(默认目标) (138) ->
(CustomBuild 目标) ->
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(271): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(319): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(271): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(319): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(271): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(319): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
我相信 TensorFlow 尚不支持 CUDA 10/Windows - 网上任何地方都没有 CUDA 10 的预编译二进制文件证实了这一点。使用 CUDA 9 重新编译，一切顺利。
我尝试在我的电脑中编译 tensorflow-1.7.0 并包含软件
cuda-10.0
cudnn-7.3
python-3.6.2,
visual studio 2017
cmake.
有错误：
不知道是怎么发生的，也不知道怎么解决
：“C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_python_build_pip_package.vcxproj”(默认目标) (1) ->
“C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\pywrap_tensorflow_internal.vcxproj”(默认目标) (2) ->
“C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels.vcxproj”(默认目标) (138) ->
(CustomBuild 目标) ->
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(271): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(319): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(271): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(319): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(271): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
C:/Users/12711/Downloads/tensorflow\tensorflow/core/kernels/reduction_gpu_kernels.cu.h(319): error : initializer not
allowed for __shared__ variable [C:\Users711\Downloads\tensorflow\tensorflow\contrib\cmake\build\tf_core_gpu_kernels
.vcxproj]
我相信 TensorFlow 尚不支持 CUDA 10/Windows - 网上任何地方都没有 CUDA 10 的预编译二进制文件证实了这一点。使用 CUDA 9 重新编译，一切顺利。