TensorFlow：解析 .d 文件时出错

Question

我用的是Ubuntu 15.10。目前我正在按照说明 here.

从源代码构建 TensorFlow

我安装成功，最后一步保存："Build your target with GPU support".

当我从我的源代码树的根部运行这个命令时：

bazel build -c opt --config=cuda //tensorflow/cc:tutorials_example_trainer

我看到以下输出：

INFO: Found 1 target...
ERROR: /home/me/tensorflow/tensorflow/core/kernels/BUILD:549:1: error while parsing .d file: /home/me/.cache/bazel/_bazel_me/38a03c514c8dc954778fc9a555bce3de/tensorflow/bazel-out/local_linux-opt/bin/tensorflow/core/kernels/_objs/colorspace_op_gpu/tensorflow/core/kernels/colorspace_op_gpu.cu.d (No such file or directory).
nvcc warning : option '--relaxed-constexpr' has been deprecated and replaced by option '--expt-relaxed-constexpr'.
In file included from third_party/gpus/cuda/include/cuda_runtime.h:76:0,
                 from <command-line>:0:
third_party/gpus/cuda/include/host_config.h:115:2: error: #error -- unsupported GNU version! gcc versions later than 4.9 are not supported!
 #error -- unsupported GNU version! gcc versions later than 4.9 are not supported!
  ^
Target //tensorflow/cc:tutorials_example_trainer failed to build
Use --verbose_failures to see the command lines of failed build steps.

有几点不明白：

为什么缺少 .d 文件？
这是什么意思？

nvcc warning : option '--relaxed-constexpr' has been deprecated and replaced by option '--expt-relaxed-constexpr'.

我的 GCC 版本是 5.2.1。这是否意味着我无法从源代码构建 TensorFlow？

感谢您的帮助。

Answer 1

考虑安装并临时选择旧版本的 gcc:

apt-get install gcc-4.8

然后更新您的备选方案：

update-alternatives --remove-all gcc
update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.8 10

编译 TensorFlow 后，您可以使用以前版本的 gcc 调用上面的命令，并使用 /usr/bin/gcc-5.x.

的正确路径

Here您可以找到更详细的答案。

TensorFlow：解析 .d 文件时出错

TensorFlow: Error while parsing .d file

ubuntu

tensorflow

bazel