No submodule power-dev

[manyiw99]

When running vision/classfication_and_detection benchmark by using docker, I cannot build the image through run_and_time.sh.

Here's the error:

[+] Building 14.9s (14/14) FINISHED                                                                                                        
 => [internal] load .dockerignore                                                                                                     0.0s
 => => transferring context: 82B                                                                                                      0.0s
 => [internal] load build definition from Dockerfile.cpu                                                                              0.0s
 => => transferring dockerfile: 1.40kB                                                                                                0.0s
 => [internal] load metadata for docker.io/library/ubuntu:16.04                                                                       1.3s
 => [ 1/11] FROM docker.io/library/ubuntu:16.04@sha256:1f1a2d56de1d604801a9671f301190704c25d604a416f59e03c04f5c6ffee0d6               0.0s
 => CACHED [ 2/11] WORKDIR /root                                                                                                      0.0s
 => CACHED [ 3/11] RUN apt-get update                                                                                                 0.0s
 => CACHED [ 4/11] RUN apt-get install -y --no-install-recommends       git       build-essential       software-properties-common    0.0s
 => CACHED [ 5/11] RUN cd /opt &&     wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-4.6.14-Linux-x86_64.sh -O minicond  0.0s
 => CACHED [ 6/11] RUN conda install pytorch-cpu torchvision-cpu -c pytorch                                                           0.0s
 => CACHED [ 7/11] RUN pip install --upgrade pip                                                                                      0.0s
 => CACHED [ 8/11] RUN pip install cmake                                                                                              0.0s
 => CACHED [ 9/11] RUN pip install future pillow onnx opencv-python-headless tensorflow onnxruntime                                   0.0s
 => CACHED [10/11] RUN pip install Cython && pip install pycocotools                                                                  0.0s
 => ERROR [11/11] RUN cd /tmp &&     git clone --recursive https://github.com/mlcommons/inference &&     cd inference/loadgen &&     13.6s
------                                                                                                                                     
 > [11/11] RUN cd /tmp &&     git clone --recursive https://github.com/mlcommons/inference &&     cd inference/loadgen &&     pip install pybind11 &&     CFLAGS="-std=c++14" python setup.py install &&     rm -rf mlperf:
#0 0.240 Cloning into 'inference'...
#0 13.48 Submodule 'language/bert/DeepLearningExamples' (https://github.com/NVIDIA/DeepLearningExamples.git) registered for path 'language/bert/DeepLearningExamples'
#0 13.50 Submodule 'third_party/pybind' (https://github.com/pybind/pybind11.git) registered for path 'third_party/pybind'
#0 13.51 fatal: no submodule mapping found in .gitmodules for path 'tools/submission/power-dev'
------
Dockerfile.cpu:39
--------------------
  38 |     RUN pip install Cython && pip install pycocotools
  39 | >>> RUN cd /tmp && \
  40 | >>>     git clone --recursive https://github.com/mlcommons/inference && \
  41 | >>>     cd inference/loadgen && \
  42 | >>>     pip install pybind11 && \
  43 | >>>     CFLAGS="-std=c++14" python setup.py install && \
  44 | >>>     rm -rf mlperf
  45 |     
--------------------
ERROR: failed to solve: process "/bin/sh -c cd /tmp &&     git clone --recursive https://github.com/mlcommons/inference &&     cd inference/loadgen &&     pip install pybind11 &&     CFLAGS=\"-std=c++14\" python setup.py install &&     rm -rf mlperf" did not complete successfully: exit code: 128
Unable to find image 'mlperf-infer-imgclassify-cpu:latest' locally
docker: Error response from daemon: pull access denied for mlperf-infer-imgclassify-cpu, repository does not exist or may require 'docker login': denied: requested access to the resource is denied.
See 'docker run --help'.

[arjunsuresh]

@manyiw99 Are you using the master branch of the inference repository with the latest changes? If so this error should not be coming.

Anyway the dockerfile for image-classification is pretty old. If you're trying the reference implementation this README can be useful.

https://github.com/mlcommons/ck/blob/master/cm-mlops/challenge/optimize-mlperf-inference-v3.1-2023/docs/generate-resnet50-submission.md

[manyiw99]

@manyiw99 Are you using the master branch of the inference repository with the latest changes? If so this error should not be coming.

Anyway the dockerfile for image-classification is pretty old. If you're trying the reference implementation this README can be useful.

https://github.com/mlcommons/ck/blob/master/cm-mlops/challenge/optimize-mlperf-inference-v3.1-2023/docs/generate-resnet50-submission.md

Yes, I used the latest master branch.

Thanks for sharing, I'll take a look at this README.

[arjunsuresh]

You're welcome. This PR should fix this error - I guess the error is coming only in older git versions.

[manyiw99]

I upgraded the git version and it didn't work. However, your new PR works for me, thank you!