这一系列基本上是属于我自己进行到了那个步骤就做到那个步骤的
由于新装了GPU (GTX750ti)和CUDA9.0、CUDNN7.1版本的软件，所以希望TensorFlow能在GPU上运行，也算上补上之前的承诺
说了下初衷，由于现在新的CUDA版本对TensorFlow的支持不好，只能采取编译源码的方式进行
所以大概分为以下几个步骤
1.安装依赖库（这部分我已经做过了，不进行介绍，可以看前边的依赖库，基本一致）
sudo apt-get install openjdk-8-jdk
jdk是bazel必须的
2.安装Git（有的就跳过这一步）
3.安装TensorFlow的build工具bazel
4.配置并编译TensorFlow源码
5.安装并配置环境变量
1.安装依赖库
2.安装Git
使用
sudo apt-get install git
git clone --recursive https://github.com/tensorflow/tensorflow
3. 安装TensorFlow的build工具bazel
这一步比较麻烦，是因为apt-get中没有bazel这个工具
因此需要到GitHub上先下载，再进行安装 下载地址是https://github.com/bazelbuild/bazel/releases
选择正确版本下载，这里序号看下TensorFlow的版本需求，具体对BAZEL的需求可以查看configure.py文件，比如我这个版本中就有这样的一段
_TF_BAZELRC_FILENAME = '.tf_configure.bazelrc' _TF_WORKSPACE_ROOT = '' _TF_BAZELRC = '' _TF_CURRENT_BAZEL_VERSION = None _TF_MIN_BAZEL_VERSION = '0.27.1' _TF_MAX_BAZEL_VERSION = '1.1.0'
每个字段的意思从字面上就可以得知，_TF_BAZELRC_FILENAME是使用bazel编译时使用的配置文件（没有特别细致的研究，https://www.cnblogs.com/shouhuxianjian/p/9416934.html里边有解释），_TF_MIN_BAZEL_VERSION = '0.27.1'是最低的bazel版本需求
使用sudo命令安装.sh文件即可
sudo chmod +x ./bazel*.sh sudo ./bazel-0.*.sh
4.配置并编译TensorFlow源码
首先是配置，可以针对自己的需求进行选择和裁剪。这一步特别麻烦，有很多选项需要选择，我的选择如下：
1 jourluohua@jour:~/tools/tensorflow$ ./configure 2 WARNING: Running Bazel server needs to be killed, because the startup options are different. 3 You have bazel 0.14.1 installed. 4 Please specify the location of python. [Default is /usr/bin/python]: 5 6 7 Found possible Python library paths: 8 /usr/local/lib/python2.7/dist-packages 9 /usr/lib/python2.7/dist-packages 10 Please input the desired Python library path to use. Default is [/usr/local/lib/python2.7/dist-packages] 11 12 Do you wish to build TensorFlow with jemalloc as malloc support? [Y/n]: Y 13 jemalloc as malloc support will be enabled for TensorFlow. 14 15 Do you wish to build TensorFlow with Google Cloud Platform support? [Y/n]: n 16 No Google Cloud Platform support will be enabled for TensorFlow. 17 18 Do you wish to build TensorFlow with Hadoop File System support? [Y/n]: n 19 No Hadoop File System support will be enabled for TensorFlow. 20 21 Do you wish to build TensorFlow with Amazon S3 File System support? [Y/n]: n 22 No Amazon S3 File System support will be enabled for TensorFlow. 23 24 Do you wish to build TensorFlow with Apache Kafka Platform support? [Y/n]: n 25 No Apache Kafka Platform support will be enabled for TensorFlow. 26 27 Do you wish to build TensorFlow with XLA JIT support? [y/N]: y 28 XLA JIT support will be enabled for TensorFlow. 29 30 Do you wish to build TensorFlow with GDR support? [y/N]: y 31 GDR support will be enabled for TensorFlow. 32 33 Do you wish to build TensorFlow with VERBS support? [y/N]: y 34 VERBS support will be enabled for TensorFlow. 35 36 Do you wish to build TensorFlow with OpenCL SYCL support? [y/N]: N 37 No OpenCL SYCL support will be enabled for TensorFlow. 38 39 Do you wish to build TensorFlow with CUDA support? [y/N]: y 40 CUDA support will be enabled for TensorFlow. 41 42 Please specify the CUDA SDK version you want to use. [Leave empty to default to CUDA 9.0]: 8 43 44 45 Please specify the location where CUDA 8.0 toolkit is installed. Refer to README.md for more details. [Default is /usr/local/cuda]: 46 47 48 Please specify the cuDNN version you want to use. [Leave empty to default to cuDNN 7.0]: 49 50 51 Please specify the location where cuDNN 7 library is installed. Refer to README.md for more details. [Default is /usr/local/cuda]: 52 53 54 Do you wish to build TensorFlow with TensorRT support? [y/N]: N 55 No TensorRT support will be enabled for TensorFlow. 56 57 Please specify the NCCL version you want to use. [Leave empty to default to NCCL 1.3]: 58 59 60 Please specify a list of comma-separated Cuda compute capabilities you want to build with. 61 You can find the compute capability of your device at: https://developer.nvidia.com/cuda-gpus. 62 Please note that each additional compute capability significantly increases your build time and binary size. [Default is: 5.0] 63 64 65 Do you want to use clang as CUDA compiler? [y/N]: N 66 nvcc will be used as CUDA compiler. 67 68 Please specify which gcc should be used by nvcc as the host compiler. [Default is /usr/bin/gcc]: 69 70 71 Do you wish to build TensorFlow with MPI support? [y/N]: N 72 No MPI support will be enabled for TensorFlow. 73 74 Please specify optimization flags to use during compilation when bazel option "--config=opt" is specified [Default is -march=native]: 75 76 77 Would you like to interactively configure ./WORKSPACE for Android builds? [y/N]: N 78 Not configuring the WORKSPACE for Android builds. 79 80 Preconfigured Bazel build configs. You can use any of the below by adding "--config=<>" to your build command. See tools/bazel.rc for more details. 81 --config=mkl # Build with MKL support. 82 --config=monolithic # Config for mostly static monolithic build. 83 Configuration finished
然后使用bazel进行编译(本步骤非常容易出问题，而且特别耗时)，这里使用 -c opt是编译release版本的，使用-c dbg是编译debug版本的
bazel build --config=opt --config=cuda //tensorflow/tools/pip_package:build_pip_package
bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
中间会遇到很多问题，这里列举一些不方便查的错误。
1）比如会遇到CXX的错误，然后具体的错误还很难排查（只显示哪个配置文件的哪一行出错，并不显示具体错误）。需要查看具体错误信息的时候，建议添加--verbose_failures选项。
2）遇到CXX的错误，（做编译的都知道，比较成熟C++的代码稳定性比较好，兼容性也比较好，移植起来也比较方便，一般不会遇到编译器和环境问题）可能是编译器gcc的版本问题，可以添加--cxxopt="-D_GLIBCXX_USE_CXX11_ABI=0"
3）遇到virtual memory exhausted: Cannot allocate memory 错误。这是因为swap分区没有设置或者swap分区容量设置太小的问题，使用free -m命令可以得知这个错误，可以使用扩展swap分区容量的方法。大概的命令如下
mkdir /home/jourluohua/swap rm -rf /home/jourluohua/swap dd if=/dev/zero of=/home/jourluohua/swap bs=1024 count=4096000
mkswap /home/jourluohua/swap
sudo swapon /home/jourluohua/swap
意思是设置4096000个1024byte大小的块，一共是4G。如果问题还是没有解决，以为bazel默认是使用多线程编译模式，可以手动添加 -j 2选项，将使用的线程固定在2
4）遇到AttributeError: 'module' object has no attribute 'IntEnum' 这个问题比较模糊，使用python -c "import enum"的时候没有错误，但是里边确实没有IntEnum的属性，查找后发现是需要安装enum34包来解决，Python不太好的一点就是各种包非常混乱，
pip install enum34 --user
5）遇到AttributeError: attribute '__doc__' of 'type' objects is not writable错误。这个问题其实挺棘手的，自身是体系结构方向，一般使用的语言也是C++，对Python不是很熟悉，也许是我的编译环境出了问题？检查查了下__doc__是Python里边注释。
先写了个小程序复现了这个问题：
#!/usr/bin/python from functools import wraps #from https://stackoverflow.com/questions/39010366/functools-wrapper-attributeerror-attribute-doc-of-type-objects-is-not def memoize(f): """ Memoization decorator for functions taking one or more arguments. Saves repeated api calls for a given value, by caching it. """ @wraps(f) class memodict(dict): """memodict""" def __init__(self, f): self.f = f def __call__(self, *args): return self[args] def __missing__(self, key): ret = self[key] = self.f(*key) return ret return memodict(f) @memoize def a(): """blah""" pass
出现了同样的错误：
Traceback (most recent call last): File "ipy.py", line 20, in <module> @memoize File "ipy.py", line 9, in memoize class memodict(dict): File "/usr/lib/python2.7/functools.py", line 33, in update_wrapper setattr(wrapper, attr, getattr(wrapped, attr)) AttributeError: attribute '__doc__' of 'type' objects is not writable
打开出问题的Python代码，原来的代码是这样
@tf_export(v1=["VariableAggregation"]) class VariableAggregation(enum.Enum): NONE = 0 SUM = 1 MEAN = 2 ONLY_FIRST_REPLICA = 3 ONLY_FIRST_TOWER = 3 # DEPRECATED def __hash__(self): return hash(self.value) # LINT.ThenChange(//tensorflow/core/framework/variable.proto) # # Note that we are currently relying on the integer values of the Python enums # matching the integer values of the proto enums. VariableAggregation.__doc__ = ( VariableAggregationV2.__doc__ + "* `ONLY_FIRST_TOWER`: Deprecated alias for `ONLY_FIRST_REPLICA`. ")
大概就是要将VariableAggregation的注释设置成VariableAggregationV2加上额外的一段"* `ONLY_FIRST_TOWER`: Deprecated alias for `ONLY_FIRST_REPLICA`. "，猜想既然不允许在class声明外做这个事情，那么直接在class中设置是否可行？
修改后的代码如下：
@tf_export(v1=["VariableAggregation"]) class VariableAggregation(enum.Enum): NONE = 0 SUM = 1 MEAN = 2 ONLY_FIRST_REPLICA = 3 ONLY_FIRST_TOWER = 3 # DEPRECATED __doc__ = (VariableAggregationV2.__doc__ + "* `ONLY_FIRST_TOWER`: Deprecated alias for `ONLY_FIRST_REPLICA`. ") def __hash__(self): return hash(self.value) # LINT.ThenChange(//tensorflow/core/framework/variable.proto) # # Note that we are currently relying on the integer values of the Python enums # matching the integer values of the proto enums. #VariableAggregation.__doc__ = ( # VariableAggregationV2.__doc__ + # "* `ONLY_FIRST_TOWER`: Deprecated alias for `ONLY_FIRST_REPLICA`. ")
6）遇到LargeZipFile: Zipfile size would require ZIP64 extensions 问题，这个问题其实很明显，就是文件太大了，在需要压缩的时候，需要配置一下ZIP64选项，而默认应该是不支持的，修改/usr/lib/python2.7/dist-packages/wheel/archive.py文件
将 zip = zipfile.ZipFile(open(zip_filename, "wb+"), "w",compression=zipfile.ZIP_DEFLATED）改成zip = zipfile.ZipFile(open(zip_filename, "wb+"), "w",compression=zipfile.ZIP_DEFLATED, allowZip64=True)就可以。
但是说实话，debug版本还是太大了，超过了zip可以压缩的大小，主要是CRC32校验那里过不去，对于我不是急需，就没有修改这里，毕竟Python2.7已经不再更新，没有努力的必要，Python3.5以上的版本这里都没有问题。
还有一些其他缺库的问题，一般都比较好搜索，就不一一列举在这里。
5.安装并配置环境变量
使用pip进行安装
$ pip install /tmp/tensorflow_pkg/tensorflow --user # with no spaces after tensorflow hit tab before hitting enter to fill in blanks
最后就是测试
import tensorflow as tf sess = tf.InteractiveSession() sess.close()
如果每一步都不报错的，TensorFlow就编译并安装成功了