Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix for pytorch2 level on import when no GPU is present #306

Merged
merged 1 commit into from
Aug 25, 2023

Conversation

jayfurmanek
Copy link
Collaborator

No description provided.

Copy link

@jithunnair-amd jithunnair-amd left a comment

Choose a reason for hiding this comment

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

LGTM, if it has been tested locally, esp. for the DeepSpeed issue that brought this up (SWDEV-412085)

@jataylo
Copy link

jataylo commented Aug 25, 2023

Verified that this commit gets us around the DeepSpeed install issue with the following Dockerfile.

# CONTEXT {'gpu_vendor': 'AMD', 'guest_os': 'UBUNTU'}
# FBGEMM requires pytorch 1.10
ARG BASE_DOCKER=compute-artifactory.amd.com:5000/rocm-plus-docker/framework/compute-rocm-dkms-no-npi-hipclang:12471_ubuntu20.04_py3.8_pytorch_release-2.0_e0e83ab

FROM $BASE_DOCKER
USER root
ENV WORKSPACE_DIR=/workspace
RUN mkdir -p $WORKSPACE_DIR
WORKDIR $WORKSPACE_DIR

# ROCm gpg key
#RUN wget -q -O - http://repo.radeon.com/rocm/rocm.gpg.key | sudo apt-key add -
#RUN apt update && apt install -y \
#    unzip \
#    jq

# add sshpass, sshfs for downloading from mlse-nas
#RUN apt-get install -y sshpass sshfs
#RUN apt-get install -y netcat

#RUN apt install net-tools

# add locale en_US.UTF-8
#RUN apt-get install -y locales
#RUN locale-gen en_US.UTF-8
#RUN pip3 install --upgrade pip

# Remove triton and install new
#RUN apt remove triton -y
RUN pip uninstall triton -y
RUN git clone https://github.com/ROCmSoftwarePlatform/triton && cd triton && git checkout 6b64ef959d41b7b229dafe7c1112b1f3753a2aa7 && cd python && python setup.py develop && cd ../..

# Install DeepSpeed (https://github.com/ROCmSoftwarePlatform/DeepSpeed/blob/master/docker/Dockerfile.rocm)
RUN pip3 install deepspeed

Merging into release/2.0 branch and will update our pinned triton commit in Pytorch release/2.0.

Copy link

@jataylo jataylo left a comment

Choose a reason for hiding this comment

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

LGTM

@jataylo jataylo merged commit 342e3f5 into release/pytorch_2.0 Aug 25, 2023
6 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants