Skip to content
This repository has been archived by the owner on Jun 5, 2023. It is now read-only.

Adding the Pytorch install options #134

Draft
wants to merge 2 commits into
base: master
Choose a base branch
from
Draft

Adding the Pytorch install options #134

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6a5d890
Deep Learning Changes tested
anubhavamd Nov 12, 2021
1ce6262
Updating all the 3 options
anubhavamd Nov 12, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
273 changes: 84 additions & 189 deletions Deep_learning/Deep-learning.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -286,21 +286,16 @@ Note: Currently, ROCm install version 3.3 is required.
Recommended: Install using published PyTorch ROCm docker image:
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note: Currently, ROCm install version 3.3 is required ?

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Change Recommended: Install using published PyTorch ROCm docker image: to
Option 1 (Recommended): Install using published PyTorch ROCm docker image:

Install or update rocm-dev on the host system: should be it's own thing since its required for all 3 options.

**************************************************************

2. Obtain docker image:

::

docker pull rocm/pytorch:rocm4.0_ubuntu18.04_py3.6_pytorch



3. Start a docker container using the downloaded image:

::

sudo docker run -it -v $HOME:/data --privileged --rm --device=/dev/kfd --device=/dev/dri --group-add video rocm/pytorch:rocm3.7_ubuntu16.04_py3.6_pytorch
2 Pull the latest public PyTorch docker image:
docker pull rocm/pytorch:latest
Optionally, you may download a specific supported configuration from https://hub.docker.com/r/rocm/pytorch.

This option provides a docker image which has PyTorch pre-installed. Users can launch the docker container and train/run deep learning models directly.

3. Start a docker container using the downloaded image:
docker run -it --privileged --device=/dev/kfd --device=/dev/dri --group-add video --ipc=host --shm-size 8G rocm/pytorch:latest
This will automatically download the image if it does not exist on the host. You can also pass -v argument to mount any data directories on to the container.

4. Confirm working installation:

::
Expand All @@ -318,182 +313,82 @@ Recommended: Install using published PyTorch ROCm docker image:
This step is optional but most PyTorch scripts will use torchvision to load models. E.g., running the pytorch examples requires torchvision.


Option 2: Install using PyTorch upstream docker file
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Option 2 title is missing

****************************************************

1. Clone PyTorch repository on the host:

::

cd ~
git clone https://github.com/pytorch/pytorch.git
cd pytorch
git submodule init
git submodule update

2. Build PyTorch docker image:

::

cd pytorch/docker/caffe2/jenkins
./build.sh py2-clang7-rocmdeb-ubuntu16.04

A message "Successfully built <image_id>" indicates a successful completion of this step.

**Note**: These steps are not tested and validated on other software versions.

3. Start a docker container using the new image:

::

sudo docker run -it -v $HOME:/data --privileged --rm --device=/dev/kfd --device=/dev/dri --group-add video <image_id>

Note: This will mount your host home directory on /data in the container.

4. Change to previous PyTorch checkout from within the running docker:

::

cd /data/pytorch

5. Build PyTorch for ROCm:

Unless you are running a gfx900/Vega10-type GPU (MI25, Vega56, Vega64,...), explicitly export the GPU architecture to build for, e.g.:
export HCC_AMDGPU_TARGET=gfx906

then
::

.jenkins/pytorch/build.sh

This will hipify the PyTorch sources first, and then compile using 4 concurrent jobs. Note, the docker image requires 16 GB of RAM.

6. Confirm working installation:

::

PYTORCH_TEST_WITH_ROCM=1 python test/run_test.py --verbose

No tests will fail if the compilation and installation is correct.

7. Install torchvision:

::

pip install torchvision

This step is optional; however, most PyTorch scripts use torchvision to load models. For example, running the pytorch examples requires torchvision.

8. Commit the container to preserve the pytorch install (from the host):

::

sudo docker commit <container_id> -m 'pytorch installed'

Option 3: Install using minimal ROCm docker file
************************************************

1. Download dockerfile based on the OS choose:
Recommend to use - Dockerfile-<OS distro>-complete to get all the ROCm Math libs installed which are required for PyTorch.

`Dockerfile <https://github.com/RadeonOpenCompute/ROCm-docker/tree/master/dev>`_

2. Build docker image:

::

sudo docker build -f ./Dockerfile-<OS distro>-complete .

The message "Successfully built <image_id>" indicates a successful completion of this step.

3. Start a docker container using the new image:

::

sudo docker run -it -v $HOME:/data --privileged --rm --device=/dev/kfd --device=/dev/dri --group-add video <image_id>

Note: This will mount your host home directory on /data in the container.

4. Clone pytorch master (on to the host):

::

cd ~
git clone https://github.com/pytorch/pytorch.git or git clone https://github.com/ROCmSoftwarePlatform/pytorch.git
cd pytorch
git submodule init
git submodule update --init --recursive'

5. Run "hipify" to prepare source code (in the container):

::

python3 tools/amd_build/build_amd.py

6. Build and install pytorch:

By default pytorch is built for all supported AMD GPU targets like gfx900/gfx906/gfx908 (MI25, MI50, MI60, MI100, ...)
This can be overwritten using
export PYTORCH_ROCM_ARCH=gfx900;gfx906;gfx908

then
::

USE_ROCM=1 MAX_JOBS=4 python3 setup.py install --user

UseMAX_JOBS=n to limit peak memory usage. If building fails try falling back to fewer jobs. 4 jobs assume available main memory of 16 GB or larger.

7. Confirm working installation:

::

PYTORCH_TEST_WITH_ROCM=1 python3 test/run_test.py --verbose

No tests will fail if the compilation and installation is correct.

8. Install torchvision:

::

pip3 install --user "git+https://github.com/pytorch/vision.git"

This step is optional. However, most PyTorch scripts will use torchvision to load models. For example, running the PyTorch examples requires torchvision.

9. Commit the container to preserve the pytorch install (from the host):

::

sudo docker commit <container_id> -m 'pyTorch installed'

PyTorch examples
*****************

1. Clone the PyTorch examples repository:

::

git clone https://github.com/pytorch/examples.git && cd examples/

2. Download pip requiremenst:

::

pip3 install -r mnist/requirements.txt

3. Run individual example: Super-resolution training and running

::

cd super_resolution/

# download dataset for training and run learning
python3 main.py --upscale_factor 3 --batchSize 4 --testBatchSize 100 --nEpochs 30 --lr 0.001

# test work super resolution effect
python3 super_resolve.py --input_image dataset/BSDS300/images/test/16077.jpg \
--model model_epoch_30.pth --output_filename out.png

4. Open `out.png` and `dataset/BSDS300/images/test/16077.jpg` files to see result
1. Obtain docker image:
docker pull rocm/pytorch:latest-base
This will download the base container, without PyTorch, to base the build upon.

2. Start a docker container using the image:
docker run -it --privileged --device=/dev/kfd --device=/dev/dri --group-add video --ipc=host --shm-size 8G rocm/pytorch:latest-base
You can also pass -v argument to mount any data directories on to the container.


3. Clone PyTorch repository:
cd ~
git clone https://github.com/pytorch/pytorch.git
cd pytorch
git submodule update --init –recursive

4. Build PyTorch for ROCm:
By default, PyTorch will build for gfx900, gfx906 and gfx908 simultaneously. To see which AMD uarch you have, run rocminfo | grep gfx (might need to install rocminfo package).
If you want to compile only for your uarch,
export PYTORCH_ROCM_ARCH=<uarch>
where <uarch> is the architecture reported by the rocminfo command.
Then build with
.jenkins/pytorch/build.sh

This will first convert PyTorch sources to be HIP compatible and then build the framework.


Option 3: Install using PyTorch upstream docker file
1. Clone PyTorch repository on the host:
cd ~
git clone https://github.com/pytorch/pytorch.git
cd pytorch
git submodule update --init --recursive
2. Build PyTorch docker image:
cd .circleci/docker
./build.sh pytorch-linux-bionic-rocm<version>-py3.6 (eg. ./build.sh pytorch-linux-bionic-rocm3.10-py3.6)
This should complete with a message "Successfully built <image_id>"
3. Clone PyTorch repository:
cd ~
git clone https://github.com/pytorch/pytorch.git
cd pytorch
git submodule update --init –recursive
4. Build PyTorch for ROCm:
By default, PyTorch will build for gfx900, gfx906 and gfx908 simultaneously. To see which AMD uarch you have, run rocminfo | grep gfx (might need to install rocminfo package).
If you want to compile only for your uarch,
export PYTORCH_ROCM_ARCH=<uarch>
where <uarch> is the architecture reported by the rocminfo command.
Then build with
.jenkins/pytorch/build.sh

This will first convert PyTorch sources to be HIP compatible and then build the framework.


Test the PyTorch installation
To validate PyTorch installation, run:

1. Test Command
cd ~ && python3 -c 'import torch' 2>/dev/null && echo "Success" || echo "Failure"
2. Running unit tests in PyTorch
Run the following command from pytorch home directory:
.jenkins/pytorch/test.sh
This runs all CI unit tests and skips as appropriate on your system based on ROCm and, e.g., single or multi GPU configuration. No tests will fail if the compilation and installation is correct. Additionally, this step will install/update a supported version of torchvision, which most PyTorch scripts use to load models.
Individual test sets can be run with:
PYTORCH_TEST_WITH_ROCM=1 python3 test/test_nn.py --verbose
where test_nn.py can be replaced with any other test set.

Try PyTorch examples
1. Clone the PyTorch examples repository:
git clone https://github.com/pytorch/examples.git
2. Run individual example: MNIST
cd examples/mnist
Follow instructions in README.md, in this case:
pip3 install -r requirements.txt
python3 main.py
3. Run individual example: Try ImageNet training
cd examples/imagenet
Follow instructions in README.md.


*********************
Expand Down