diff options
Diffstat (limited to 'tensorflow/contrib/makefile/README.md')
-rw-r--r-- | tensorflow/contrib/makefile/README.md | 29 |
1 files changed, 18 insertions, 11 deletions
diff --git a/tensorflow/contrib/makefile/README.md b/tensorflow/contrib/makefile/README.md index f82b70187c..f7f46fca5c 100644 --- a/tensorflow/contrib/makefile/README.md +++ b/tensorflow/contrib/makefile/README.md @@ -42,7 +42,7 @@ at `tensorflow/contrib/makefile/gen/bin/benchmark`. To run the executable, use: tensorflow/contrib/makefile/gen/bin/benchmark --graph=tensorflow_inception_graph.pb ``` -You should download the example graph from [http://download.tensorflow.org/models/image/imagenet/inception-2015-12-05.tgz](http://download.tensorflow.org/models/image/imagenet/inception-2015-12-05.tgz). +You should download the example graph from [https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip](https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip). ## Supported Systems @@ -132,24 +132,31 @@ static library in a simple app. ## Raspberry Pi -The easiest way to build for the Raspberry Pi is to cross-compile from Linux. -To use this makefile to do that, you first need to install the right version of -the compiler to target the Pi, using a command like this on your Linux machine: +Building on the Raspberry Pi is similar to a normal Linux system, though we +recommend starting by compiling and installing protobuf: ```bash -sudo apt-get install g++-arm-linux-gnueabihf +cd tensorflow/contrib/makefile/downloads/protobuf/ +./autogen.sh +./configure +make +sudo make install +cd ../../../../.. ``` -After that, run `tensorflow/contrib/makefile/compile_pi_protobuf.sh` to build a -version of the protobuf library aimed at the Pi. Then you should be able to run: +Once that's done, you can use make to build the library and example: ```bash -make -f tensorflow/contrib/makefile/Makefile TARGET=PI +make -f tensorflow/contrib/makefile/Makefile HOST_OS=PI TARGET=PI OPTFLAGS="-Os" ``` -This will build the static library, and the example benchmark executable. You -can then copy the `tensorflow/contrib/makefile/gen/bin/benchmark` program over -to your Raspberry Pi, and run it there. +If you're only interested in building for Raspberry Pi's 2 and 3, you can supply +some extra optimization flags to give you code that will run faster: + +```bash +make -f tensorflow/contrib/makefile/Makefile HOST_OS=PI TARGET=PI \ +OPTFLAGS="-Os -mfpu=neon-vfpv4 -funsafe-math-optimizations -ftree-vectorize" +``` ## Dependencies |