| import segyio | |
| filename = 'f3.segy' | |
| src_data = segyio.tools.cube(filename) | |
| k1,k2,k3 = 300, 500, 400 | |
| factor = 125 | |
| il_s = k1 - factor | |
| il_e = k1 + factor |
| #https://github.intel.com/CloudPassage/AWS-Halo-Deployment-Script | |
| wget https://www.dropbox.com/s/imh6zsh8nrpubzd/AWS-Halo-Deployment-Script-master.zip | |
| unzip AWS-Halo-Deployment-Script-master.zip | |
| cd AWS-Halo-Deployment-Script-master/ | |
| sudo python HaloAgent.py |
| # Launch a AMZ DL AMI (Ubuntu) | |
| #Install Bazel with --user, it will be installed at /home/ubuntu/.bazel | |
| wget https://github.com/bazelbuild/bazel/releases/download/0.13.1/bazel-0.13.1-installer-linux-x86_64.sh | |
| chmod +x bazel-0.13.1-installer-linux-x86_64.sh | |
| ./bazel-0.13.1-installer-linux-x86_64.sh --user | |
| source /home/ubuntu/.bazel/bin/bazel-complete.bash | |
| #Create and activate a virtual envirnoment. | |
| conda create -n tf1.8_mkl-master_build python=3.6 | |
| source activate tf1.8_mkl-master_build |
| #Example Inference benchmark on Amazon DL AMI. | |
| #Launch a DL AMI | |
| #Create an virtual envirnoment. | |
| conda create -n tf1.8_mkl-master python=3.6 | |
| source activate tf1.8_mkl-master | |
| #Download the tf1.8 wheel built with MKLDNN master branch on 05-31-2018 | |
| wget https://www.dropbox.com/s/y812q7zpdy4pjed/tensorflow-1.8.0-cp36-cp36m-linux_x86_64.whl | |
| #pip install the wheel |
| # !/bin/bash | |
| # Run TensorFlow's tf_cnn benchmarks in tensorflow_p36 virtualenv | |
| # Activate TensorFlow virtual environment | |
| #source activate tensorflow_p36 | |
| # Clone benchmark scripts | |
| git clone https://github.com/tensorflow/benchmarks.git | |
| cd benchmarks/scripts/tf_cnn_benchmarks/ |
Docker install instructions: https://github.com/ravi9/misc-readmes/blob/master/install-docker-centos.md
docker pull centos/devtoolset-6-toolchain-centos7
#Login into the docker with with USER 0 with sudo permissions.
docker run -it --user 0 centos/devtoolset-6-toolchain-centos7:latest /bin/bash
| #Quick benchmarking resnet50 with MKL and NO MKL on TF | |
| #Following are steps: | |
| # Create MKL virtual envirnoment | |
| # pip install a TF1.8 MKL Wheel | |
| # git clone TF_CNN benchmarks. | |
| # run resnet50 inference benchmark. | |
| # | |
| # Create NO-MKL virtual envirnoment | |
| # pip install a TF1.8 Wheel this will install eigen based TF |
| #Install requirements. | |
| pip install virtualenv | |
| virtualenv --python=/usr/bin/python2.7 keras_tracing | |
| source keras_tracing/bin/activate | |
| cd keras_tracing/ | |
| pip install keras | |
| pip install -i https://pypi.anaconda.org/intel/simple tensorflow | |
| #Clone mnist sample code | |
| git clone https://github.com/ravi9/keras-trace-example.git |
| #Git clone NMT | |
| git clone https://github.com/tensorflow/nmt.git | |
| #Download Dataset | |
| cd nmt | |
| nmt/scripts/download_iwslt15.sh /home/ubuntu/nmt_data | |
| #Run training for few steps to get a model, so that we can use it for inference benchmarking. | |
| python -m nmt.nmt \ | |
| --src=vi --tgt=en \ |
| #!/bin/bash | |
| #nmt_infer_batchsize.sh | |
| ### | |
| #Prereq: | |
| #sudo apt install -y moreutils jq | |
| ### | |
| echo -e "\nBZ WPS \n" | |
| #set -x |