TensorRT Initial commit

[azhurkevich]

trafficstars

Please add a meaningful description for your change here Integration of TensorRT into Apache Beam.

Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:

• [x] Choose reviewer(s) and mention them in a comment (R: @tvalentyn @AnandInguva @ryanthompson591 @yeandy).
• [ ] Mention the appropriate issue in your description (for example: addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, comment fixes #<ISSUE NUMBER> instead.
• [ ] Update CHANGES.md with noteworthy changes.
• [ ] If this contribution is large, please file an Apache Individual Contributor License Agreement.

See the Contributor Guide for more tips on how to make review process smoother.

To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI.

[codecov[bot]]

Codecov Report

Merging #22131 (2c417db) into master (48513ad) will decrease coverage by 0.19%. The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master   #22131      +/-   ##
==========================================
- Coverage   74.18%   73.99%   -0.20%     
==========================================
  Files         706      714       +8     
  Lines       93229    93945     +716     
==========================================
+ Hits        69159    69510     +351     
- Misses      22802    23167     +365     
  Partials     1268     1268              

Flag	Coverage Δ
python	83.18% <0.00%> (-0.38%)	:arrow_down:

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...am/examples/inference/tensorrt_object_detection.py	0.00% <0.00%> (ø)
...hon/apache_beam/ml/inference/tensorrt_inference.py	0.00% <0.00%> (ø)
sdks/python/apache_beam/typehints/__init__.py	77.77% <0.00%> (-22.23%)	:arrow_down:
...ks/python/apache_beam/runners/worker/statecache.py	89.69% <0.00%> (-6.47%)	:arrow_down:
.../python/apache_beam/testing/test_stream_service.py	88.09% <0.00%> (-4.77%)	:arrow_down:
...dks/python/apache_beam/metrics/monitoring_infos.py	92.50% <0.00%> (-4.50%)	:arrow_down:
...examples/inference/sklearn_mnist_classification.py	43.75% <0.00%> (-3.75%)	:arrow_down:
sdks/python/apache_beam/internal/pickler.py	92.00% <0.00%> (-3.46%)	:arrow_down:
...python/apache_beam/runners/worker/worker_status.py	76.66% <0.00%> (-3.05%)	:arrow_down:
sdks/python/apache_beam/utils/interactive_utils.py	95.12% <0.00%> (-2.44%)	:arrow_down:
... and 51 more

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

[github-actions[bot]]

Assigning reviewers. If you would like to opt out of this review, comment assign to next reviewer:

R: @ryanthompson591 for label python.

Available commands:

• stop reviewer notifications - opt out of the automated review tooling
• remind me after tests pass - tag the comment author after tests pass
• waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

[yeandy]

R: @yeandy @AnandInguva @tvalentyn

[github-actions[bot]]

Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control

[azhurkevich]

@yeandy I think this will be it for the first initial TensorRT integration from my side. Moving multi-input models into future integration.

[yeandy]

Sounds good, thanks!

[azhurkevich]

@yeandy I'll look into what you've mentioned tomorrow, will try to address the issues as soon as I figure them out.

[yeandy]

Actually, I have been experimenting around with the docs stuff in your branch. I can do a quick fix right now.

[yeandy]

The Sphinx docs generation requires module imports import tensorrt as trt and from cuda import cuda. I think our script just needs to detect that those modules exist in tensorrt_inference.py via the import statements. @azhurkevich Want to confirm before I pursue a different route: To your knowledge, is there a way to install these two packages without having GPUs, drivers, toolkit, etc. set up? (I was trying pip installation method locally, but ran into issues)

[pranavm-nvidia]

@yeandy With Sphinx, you can specify modules to mock in your conf.py (example). With that, you won't need to have TensorRT installed.

[yeandy]

Oh, great to know! Thanks @pranavm-nvidia, I'll check that out

[tvalentyn]

Just a heads-up, Andy is OOO atm, so his responses may be delayed till next week.

[tvalentyn]

Run PythonLint PreCommit

[azhurkevich]

@ryanthompson591 My bad, forgot about implementation. Pushed now, thanks!

[AnandInguva]

Run Python 3.8 PostCommit

[yeandy]

Run Python 3.8 PostCommit

[tvalentyn]

I took a closer look, see suggestions. I haven't tested it.

Pushed the suggestions, let's see if the tests pass.

[AnandInguva]

Run Python 3.8 PostCommit

[damccorm]

Run Python PreCommit

[damccorm]

Run Python PreCommit

[yeandy]

Good job everyone, and thanks @damccorm for the merge!