CPU/GPU interoperability POC #4874

viclafargue · 2022-08-23T08:15:11Z

No description provided.

wphicks

This looks really great! I think it covers just about everything that's needed. Only thing I don't love is the name of interoperability.py, since I'm not sure it would be immediately clear to new developers what that refers to. Maybe device_selection.py or something else instead?

…tor (rapidsai#4871) Authors: - Dante Gama Dessavre (https://github.com/dantegd) Approvers: - William Hicks (https://github.com/wphicks) - Corey J. Nolet (https://github.com/cjnolet) - Victor Lafargue (https://github.com/viclafargue) URL: rapidsai#4871

…sai#4862) Removes possibility of another projects RAPIDS.cmake being used, and removes need to always download a version. Authors: - Robert Maynard (https://github.com/robertmaynard) Approvers: - Dante Gama Dessavre (https://github.com/dantegd) URL: rapidsai#4862

For housekeeping, this PR removes unused cuDF imports across a variety of files. It does not refactor any code that currently relies on cuDF. Authors: - Nick Becker (https://github.com/beckernick) Approvers: - Dante Gama Dessavre (https://github.com/dantegd) URL: rapidsai#4873

Import treelite models into FIL in a different precision. - e.g. load float64 treelite models as a float32 FIL model, or vice versa Authors: - Andy Adinets (https://github.com/canonizer) - William Hicks (https://github.com/wphicks) Approvers: - Philip Hyunsu Cho (https://github.com/hcho3) - William Hicks (https://github.com/wphicks) URL: rapidsai#4839

- [x] All points distance membership vector - [x] All points outlier membership vector - [x] All points probability in some cluster - [x] All points membership vector - [x] Tests Authors: - Tarang Jain (https://github.com/tarang-jain) Approvers: - Corey J. Nolet (https://github.com/cjnolet) URL: rapidsai#4800

beckernick · 2022-09-07T20:58:03Z

Gave this a test locally. It's really smooth! Quick comment:

When dispatching linear regression to CPU, the weights/bias attributes are not propagated back up to the cuML estimator -- they're accessible only via the underlying sk_model_. Ideally, cuML users won't need to think about the distinction between PGU and CPU dispatched models. They can access attributes in the same way regardless of hardware.

import cuml
from cuml.common.device_selection import using_device_type, set_global_device_type
from sklearn.datasets import make_regression

X, y = make_regression(
    n_samples=100000,
    noise=50,
    n_features=100,
)

set_global_device_type("cpu")

clf = cuml.linear_model.LinearRegression()
clf.fit(X,y)
print(clf.score(X, y))
print(clf.coef_, clf.intercept_)
0.9451533686273299
None None

python/cuml/common/cuda.pyx

dantegd · 2022-10-05T16:36:55Z

rerun tests

cjnolet · 2022-10-06T16:44:29Z

I'm waiting on the raft packages to deploy to the rapidsai-nightly channel and then we should be able to rerun this PR.

viclafargue · 2022-10-06T16:56:30Z

I'm waiting on the raft packages to deploy to the rapidsai-nightly channel and then we should be able to rerun this PR.

Should we remove the latest commit?

cjnolet · 2022-10-06T17:42:17Z

Should we remove the latest commit?

Yes. Though it might be easier to revert the commit so you don't have to rebase / force push.

This reverts commit 09db802.

cjnolet · 2022-10-06T19:56:32Z

JFYI: We're still waiting for pylibraft / raft-dask to be deployed to conda, which could take another 25-40mins. I reverted the last commit in the meantime.

cjnolet · 2022-10-06T20:47:44Z

rerun tests

viclafargue · 2022-10-07T11:08:56Z

rerun tests

codecov-commenter · 2022-10-07T12:25:09Z

Codecov Report

Base: 79.26% // Head: 79.30% // Increases project coverage by +0.04% 🎉

Coverage data is based on head (495c676) compared to base (d65121a).
Patch coverage: 95.12% of modified lines in pull request are covered.

Additional details and impacted files

@@               Coverage Diff                @@
##           branch-22.10    #4874      +/-   ##
================================================
+ Coverage         79.26%   79.30%   +0.04%     
================================================
  Files               181      183       +2     
  Lines             11540    11618      +78     
================================================
+ Hits               9147     9214      +67     
- Misses             2393     2404      +11

Flag	Coverage Δ
dask	`46.10% <58.53%> (+0.07%)`	⬆️
non-dask	`68.77% <93.90%> (+0.11%)`	⬆️

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

Impacted Files	Coverage Δ
python/cuml/common/array_descriptor.py	`100.00% <ø> (ø)`
python/cuml/common/input_utils.py	`93.85% <50.00%> (-0.39%)`	⬇️
python/cuml/internals/global_settings.py	`91.89% <86.36%> (-8.11%)`	⬇️
python/cuml/common/__init__.py	`100.00% <100.00%> (ø)`
python/cuml/common/device_selection.py	`100.00% <100.00%> (ø)`
python/cuml/common/memory_utils.py	`82.35% <100.00%> (+2.47%)`	⬆️
python/cuml/dask/common/base.py	`94.44% <100.00%> (+0.03%)`	⬆️
python/cuml/experimental/common/__init__.py	`100.00% <100.00%> (ø)`
python/cuml/explainer/common.py	`88.57% <100.00%> (+0.16%)`	⬆️
python/cuml/testing/utils.py	`89.19% <100.00%> (+0.07%)`	⬆️
... and 1 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

cjnolet · 2022-10-07T12:26:55Z

@viclafargue webare past code freeze for 22.10 and decided to push this to 22.12. Can you base this ok the 22.12 branch?

dantegd · 2022-10-17T18:25:28Z

rerun tests

dantegd · 2022-10-18T19:09:33Z

@gpucibot merge

Add device and memory type settings

06df028

viclafargue requested a review from a team as a code owner August 23, 2022 08:15

github-actions bot added the Cython / Python Cython or Python issue label Aug 23, 2022

wphicks reviewed Aug 23, 2022

View reviewed changes

viclafargue added 3 commits August 24, 2022 15:05

Apply changes

ba70967

Add pytests

d19a6ef

CPU/GPU dispatch according to device_type

687bf73

dantegd added the 2 - In Progress Currenty a work in progress label Aug 30, 2022

viclafargue and others added 9 commits September 1, 2022 14:01

Testing different input types

f120ac5

Add code doc

1556de8

Remove None of acceptable values for device_type and memory_type

62f4ea7

merging

0cdafd5

github-actions bot added CMake CUDA/C++ labels Sep 2, 2022

merging and style fix

ff6d60a

divyegala requested review from a team as code owners September 7, 2022 18:48

github-actions bot removed CMake CUDA/C++ labels Sep 7, 2022

divyegala added feature request New feature or request non-breaking Non-breaking change labels Sep 7, 2022

beckernick reviewed Sep 7, 2022

View reviewed changes

python/cuml/common/cuda.pyx Show resolved Hide resolved

Fix import

bfe238f

viclafargue force-pushed the cpu-gpu-interop branch from 1836994 to bfe238f Compare October 4, 2022 14:00

Fix tests

5c03489

viclafargue force-pushed the cpu-gpu-interop branch from 4f7917c to 5c03489 Compare October 5, 2022 11:54

Using raft::KeyValuePair instead of cub::KeyValuePair

09db802

github-actions bot added the CUDA/C++ label Oct 5, 2022

Revert "Using raft::KeyValuePair instead of cub::KeyValuePair"

0dbe8a4

This reverts commit 09db802.

github-actions bot removed the CUDA/C++ label Oct 6, 2022

Fix _check_internal_model

495c676

viclafargue changed the base branch from branch-22.10 to branch-22.12 October 7, 2022 12:37

viclafargue force-pushed the cpu-gpu-interop branch from d632c60 to 495c676 Compare October 7, 2022 13:15

viclafargue added 2 commits October 7, 2022 15:29

Merge branch 'branch-22.10' into cpu-gpu-interop

b8c2e97

Fix pydoc issue

243b23e

rapids-bot bot merged commit cd23871 into rapidsai:branch-22.12 Oct 18, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

CPU/GPU interoperability POC #4874

CPU/GPU interoperability POC #4874

viclafargue commented Aug 23, 2022

wphicks left a comment

beckernick commented Sep 7, 2022

dantegd commented Oct 5, 2022

cjnolet commented Oct 6, 2022

viclafargue commented Oct 6, 2022

cjnolet commented Oct 6, 2022

cjnolet commented Oct 6, 2022

cjnolet commented Oct 6, 2022

viclafargue commented Oct 7, 2022

codecov-commenter commented Oct 7, 2022

cjnolet commented Oct 7, 2022

dantegd commented Oct 17, 2022

dantegd commented Oct 18, 2022

CPU/GPU interoperability POC #4874

CPU/GPU interoperability POC #4874

Conversation

viclafargue commented Aug 23, 2022

wphicks left a comment

Choose a reason for hiding this comment

beckernick commented Sep 7, 2022

dantegd commented Oct 5, 2022

cjnolet commented Oct 6, 2022

viclafargue commented Oct 6, 2022

cjnolet commented Oct 6, 2022

cjnolet commented Oct 6, 2022

cjnolet commented Oct 6, 2022

viclafargue commented Oct 7, 2022

codecov-commenter commented Oct 7, 2022

Codecov Report

cjnolet commented Oct 7, 2022

dantegd commented Oct 17, 2022

dantegd commented Oct 18, 2022