[REVIEW] upgrade to RAPIDS 0.11

NVIDIA / fsi-samples

A collection of open-source GPU accelerated Python tools and examples for quantitative analyst tasks and leverages RAPIDS AI project, Numba, cuDF, and Dask.

271 stars 115 forks source link

[REVIEW] upgrade to RAPIDS 0.11 #76

Closed yidong72 closed 4 years ago

yidong72 commented 4 years ago

Since RAPIDS 0.11 is released, move the changes from PR to this PR. I didn't include

RUN echo "conda 4.7.12" > /opt/conda/conda-meta/pinned \
    && conda update -n base -c defaults conda -y

Not convinced it solved the problem.

miguelusque commented 4 years ago

Hi Yi,

The new RAPIDS container (v0.11) already includes conda v4.7.12, so you are right, that change is not needed anymore.

Thank you for taking care of this PR.

Regards, Miguel

yidong72 commented 4 years ago

Hi Yi,

The new RAPIDS container (v0.11) already includes conda v4.7.12, so you are right, that change is not needed anymore.

Thank you for taking care of this PR.

Regards, Miguel

Thanks for the clarification of this.

yidong72 commented 4 years ago

Added the profile argument to help optimize the performance. Now we can evaluate the graph by:

o_gpu = task_graph.run(
            outputs=outlist,
            replace=replace_spec, profile=True)

yidong72 commented 4 years ago

I did all the changes as you suggested. please review it again.

yidong72 commented 4 years ago

the CUDA version tag is fixed