sono-bfio

export NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5
export NVIDIA_GPGKEY_FPR=ae09fe4bbd223a84b2ccfce3f60f4b3d7fa2af80
apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub
apt-key adv --export --no-emit-version -a $NVIDIA_GPGKEY_FPR | tail -n +5 > cudasign.pub
echo "$NVIDIA_GPGKEY_SUM  cudasign.pub" | sha256sum -c --strict - && rm cudasign.pub
echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list

apt-get update && apt-get install -y --no-install-recommends cuda-8-0*

sono-bfio

# Ubuntu 16 Verified
apt-get update && apt-get install -y curl apt-transport-https
curl -o bootstrap-salt.sh -L https://bootstrap.saltstack.com
sh bootstrap-salt.sh -P git v2016.11.4

echo 'file_client: local'                 > /etc/salt/minion
echo ''                                   >> /etc/salt/minion
echo ''                                   >> /etc/salt/minion
echo 'log_level: info'                    >> /etc/salt/minion
echo 'log_level_logfile: info'            >> /etc/salt/minion

sono-bfio

Nvidia Repo Setup

NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
NVIDIA_GPGKEY_FPR=ae09fe4bbd223a84b2ccfce3f60f4b3d7fa2af80 && \
apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1404/x86_64/7fa2af80.pub && \
apt-key adv --export --no-emit-version -a $NVIDIA_GPGKEY_FPR | tail -n +2 > cudasign.pub && \
echo "$NVIDIA_GPGKEY_SUM  cudasign.pub" | sha256sum -c --strict - && rm cudasign.pub && \
echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1404/x86_64 /" > /etc/apt/sources.list.d/cuda.list

sono-bfio

Install PyCuda

Option 1:

# As the root user
sudo su -
pip install pycuda
exit

sono-bfio

On the G2.8X we have a total of 8 devices. To expose them to Torch set the environment variable.

• Note: can't scale much past 8 GPUs due to luajit 1G memory limit *

export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
th train-on-mnist-with-gpu.lua 

LUA Script to Train the Model below:

sono-bfio

from __future__ import division, print_function, absolute_import

# Import tflearn and some helpers
import tflearn
from tflearn.data_utils import shuffle
from tflearn.layers.core import input_data, dropout, fully_connected
from tflearn.layers.conv import conv_2d, max_pool_2d
from tflearn.layers.estimator import regression
from tflearn.data_preprocessing import ImagePreprocessing
from tflearn.data_augmentation import ImageAugmentation

sono-bfio

Upgrading the GPU nodes

1. Launch Bitfusion Boost cluster

2. Log into the client node

$ ssh -l ubuntu -i path/to/your_key.pem {IP}

sono-bfio

Upgrading the GPU nodes

1. Launch Bitfusion Boost cluster

2. Log into the client node

$ ssh -l ubuntu -i path/to/your_key.pem {IP}

sono-bfio

from __future__ import division, print_function, absolute_import

# Import tflearn and some helpers
import tflearn
from tflearn.data_utils import shuffle
from tflearn.layers.core import input_data, dropout, fully_connected
from tflearn.layers.conv import conv_2d, max_pool_2d
from tflearn.layers.estimator import regression
from tflearn.data_preprocessing import ImagePreprocessing
from tflearn.data_augmentation import ImageAugmentation

sono-bfio

Installing tf learn - python 2.7

/usr/local/bin/pip2 tflearn

Install tf learn - python 3.4

/usr/local/bin/pip3 tflearn