WIP: [DO NOT MERGE] introduce libcuml wheels #6199

jameslamb · 2024-12-30T23:31:32Z

Replaces #6006, contributes to rapidsai/build-planning#33.

Proposes packaging libcuml as a wheel, which is then re-used by cuml-cu{11,12} wheels.

Notes for Reviewers

If you see this note, that means this is not ready for review.

Benefits of these changes

smaller wheels (see "Size Changes" below)
faster compile times
- no more re-compiling RAFT, thanks to introduce libraft wheels raft#2531
less use of CI resources (only compiling once per CPU architecture / CUDA versions, instead of once per those + Python minor version)
other benefits mentioned in Support dynamic linking between RAPIDS wheels build-planning#33

Wheel contents

libcuml:

libcuml++.so (shared library) and its headers
libcumlprims_mg.so (shared library) and its headers
other vendored dependencies (CCCL, fmt)

cuml:

cuml Python / Cython code and compiled Cython extensions

Dependency Flows

In short.... libcuml contains libcuml.so and libcumlprims_mg.so dynamic libraries and the headers to link against them.

Anything that needs to link against cuML at build time pulls in libcugraph wheels as a build dependency.
Anything that needs cuML's symbols at runtime pulls it in as a runtime dependency, and calls libcuml.load_library().

For more details and some flowcharts, see rapidsai/build-planning#33 (comment)

Size changes (CUDA 12, Python 3.12, x86_64)

wheel	num files (before)	num files (this PR)	size (before)	size (these PRs)
`libcuml`	---	---	---	---
`cuml`	442	---	517M	---
TOTAL	442	---	517M	---

NOTES: size = compressed, "before" = 2025-01-13 nightlies

how I calculated those (click me)

nightly commit = 7c715c4
PR = this PR

docker run \
    --rm \
    --network host \
    --env RAPIDS_NIGHTLY_DATE=2025-01-13 \
    --env CUML_NIGHTLY_SHA=7c715c494dff71274d0fdec774bdee12a7e78827 \
    --env CUML_PR="pull-request/6199" \
    --env CUML_PR_SHA="7c715c494dff71274d0fdec774bdee12a7e78827" \
    --env RAPIDS_PY_CUDA_SUFFIX=cu12 \
    --env WHEEL_DIR_BEFORE=/tmp/wheels-before \
    --env WHEEL_DIR_AFTER=/tmp/wheels-after \
    -it rapidsai/ci-wheel:cuda12.5.1-rockylinux8-py3.12 \
    bash

# --- nightly wheels --- #
mkdir -p ./wheels-before

export RAPIDS_BUILD_TYPE=branch
export RAPIDS_REF_NAME="branch-25.02"

# cuml
RAPIDS_PY_WHEEL_NAME="cuml_${RAPIDS_PY_CUDA_SUFFIX}" \
RAPIDS_REPOSITORY=rapidsai/cuml \
RAPIDS_SHA=${CUML_NIGHTLY_SHA} \
    rapids-download-wheels-from-s3 python ./wheels-before

# --- wheels from CI --- #
mkdir -p ./wheels-after

export RAPIDS_BUILD_TYPE="pull-request"

# libcuml
RAPIDS_PY_WHEEL_NAME="libcuml_${RAPIDS_PY_CUDA_SUFFIX}" \
RAPIDS_REPOSITORY=rapidsai/cuml \
RAPIDS_REF_NAME="${CUML_PR}" \
RAPIDS_SHA="${CUML_PR_SHA}" \
    rapids-download-wheels-from-s3 cpp ./wheels-after

# cuml
RAPIDS_PY_WHEEL_NAME="cuml_${RAPIDS_PY_CUDA_SUFFIX}" \
RAPIDS_REPOSITORY=rapidsai/cuml \
RAPIDS_REF_NAME="${CUML_PR}" \
RAPIDS_SHA="${CUML_PR_SHA}" \
    rapids-download-wheels-from-s3 python ./wheels-after

pip install pydistcheck
pydistcheck \
    --inspect \
    --select 'distro-too-large-compressed' \
    ./wheels-before/*.whl \
| grep -E '^checking|files: | compressed' \
> ./before.txt

# get more exact sizes
du -sh ./wheels-before/*

pydistcheck \
    --inspect \
    --select 'distro-too-large-compressed' \
    ./wheels-after/*.whl \
| grep -E '^checking|files: | compressed' \
> ./after.txt

# get more exact sizes
du -sh ./wheels-after/*

How I tested this

These other PRs:

(TODO: add devcontainers PR)

…wheels

…ges (#6217) Follow-up to #6190. Proposes some miscellaneous packaging cleanup: * declares `cuml-cu{11,12}` wheels' runtime dependency on `cuda-python` - *as a result of stuff like this: https://github.com/rapidsai/cuml/blob/bfd2e220d3adf5d8c6b76dc90e3d1275054f32d5/python/cuml/cuml/svm/linear.pyx#L40-L43* *~ adds `raft_log.txt` to `.gitignore`~ * adds CMake option `CUML_USE_RAFT_STATIC` - *to provide a default for this: https://github.com/rapidsai/cuml/blob/bfd2e220d3adf5d8c6b76dc90e3d1275054f32d5/cpp/CMakeLists.txt#L600* * defines `BUILD_CAGRA_HNSWLIB OFF` in `get_cuvs.cmake` - *as is done for RAFT: https://github.com/rapidsai/cuml/blob/bfd2e220d3adf5d8c6b76dc90e3d1275054f32d5/cpp/cmake/thirdparty/get_raft.cmake#L58* - *cuML doesn't need the CAGRA stuff from cuVS, as far as I can tell* - *this is `ON` by default in cuVS, so this change saves a bit of build time and size: https://github.com/rapidsai/cuvs/blob/1e548f8c3a773452ce69556f4db72fc712efae02/cpp/CMakeLists.txt#L58* * explicitly passing package type to `rapids-download-wheels-from-s3` in CI scripts ## Notes for Reviewers These changes are useful independently, but will also make the `libcuml` wheels PR (#6199) a bit smaller and easier to review. Authors: - James Lamb (https://github.com/jameslamb) Approvers: - Vyas Ramasubramani (https://github.com/vyasr) URL: #6217

…wheels

Replaces #2306, contributes to rapidsai/build-planning#33. Proposes packaging `libraft` as a wheel, which is then re-used by: * `pylibraft-cu{11,12}` and `raft-cu{11,12}` (this PR) * `libcugraph-cu{11,12}`, `pylibcugraph-cu{11,12}`, and `cugraph-cu{11,12}` in rapidsai/cugraph#4804 * `libcuml-cu{11,12}` and `cuml-cu{11,12}` in rapidsai/cuml#6199 As part of this, also proposes: * introducing a new CMake option, `RAFT_COMPILE_DYNAMIC_ONLY`, to allow building/installing only the dynamic shared library (i.e. skipping the static library) * enforcing `rapids-cmake`'s preferred CMake style (#2531 (comment)) * making wheel-building CI jobs always depend on other wheel-building CI jobs, not tests or `*-publish` (to reduce end-to-end CI time) ## Notes for Reviewers ### Benefits of these changes * smaller wheels (see "Size Changes" below) * faster compile times (no more re-compiling RAFT in cuGraph and cuML CI) * other benefits mentioned in rapidsai/build-planning#33 ### Wheel contents `libraft`: * `libraft.so` (shared library) * RAFT headers * vendored dependencies (`fmt`, CCCL, `cuco`, `cute`, `cutlass`) `pylibraft`: * `pylibraft` Python / Cython code and compiled Cython extensions `raft-dask`: * `raft-dask` Python / Cython code and compiled Cython extension ### Dependency Flows In short.... `libraft` contains a `libraft.so` dynamic library and the headers to link against it. * Anything that needs to link against RAFT at build time pulls in `libraft` wheels as a build dependency. * Anything that needs RAFT's symbols at runtime pulls it in as a runtime dependency, and calls `libraft.load_library()`. For more details and some flowcharts, see rapidsai/build-planning#33 (comment) ### Size changes (CUDA 12, Python 3.12, x86_64) | wheel | num files (before) | num files (these PRs) | size (before) | size (these PRs) | |:---------------:|------------------:|-----------------:|--------------:|-------------:| | `libraft`. | --- | 3169 | --- | 19M | | `pylibraft` | 64 | 63 | 11M | 1M | | `raft-dask` | 29 | 28 | 188M | 188M | | `libcugraph` | --- | 1762 | --- | 903M | | `pylibcugraph` | 190 | 187 | 901M | 2M | | `cugraph` | 315 | 313 | 899M | 3.0M | | `libcuml` | --- | 1766 | --- | 289M | | `cuml` | 442 | --- | 517M | --- | |**TOTAL** | **1,040** | **7,268** | **2,516M** | **1,405M** | *NOTES: size = compressed, "before" = 2025-01-13 nightlies* <details><summary>how I calculated those (click me)</summary> * `cugraph`: nightly commit = rapidsai/cugraph@8507cbf, PR = rapidsai/cugraph#4804 * `cuml`: nightly commit = rapidsai/cuml@7c715c4, PR = rapidsai/cuml#6199 * `raft`: nightly commit = 1b62c41, PR = this PR ```shell docker run \ --rm \ --network host \ --env RAPIDS_NIGHTLY_DATE=2025-01-13 \ --env CUGRAPH_NIGHTLY_SHA=8507cbf63db2f349136b266d3e6e787b189f45a0 \ --env CUGRAPH_PR="pull-request/4804" \ --env CUGRAPH_PR_SHA="2ef32eaa006a84c0bd16220bb8e8af34198fbee8" \ --env CUML_NIGHTLY_SHA=7c715c494dff71274d0fdec774bdee12a7e78827 \ --env CUML_PR="pull-request/6199" \ --env CUML_PR_SHA="2ef32eaa006a84c0bd16220bb8e8af34198fbee8" \ --env RAFT_NIGHTLY_SHA=1b62c4117a35b11ce3c830daae248e32ebf75e3f \ --env RAFT_PR="pull-request/2531" \ --env RAFT_PR_SHA="0d6597b08919f2aae8ac268f1a68d6a8fe5beb4e" \ --env RAPIDS_PY_CUDA_SUFFIX=cu12 \ --env WHEEL_DIR_BEFORE=/tmp/wheels-before \ --env WHEEL_DIR_AFTER=/tmp/wheels-after \ -it rapidsai/ci-wheel:cuda12.5.1-rockylinux8-py3.12 \ bash # --- nightly wheels --- # mkdir -p ./wheels-before export RAPIDS_BUILD_TYPE=branch export RAPIDS_REF_NAME="branch-25.02" # pylibraft RAPIDS_PY_WHEEL_NAME="pylibraft_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/raft \ RAPIDS_SHA=${RAFT_NIGHTLY_SHA} \ rapids-download-wheels-from-s3 python ./wheels-before # raft-dask RAPIDS_PY_WHEEL_NAME="raft_dask_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/raft \ RAPIDS_SHA=${RAFT_NIGHTLY_SHA} \ rapids-download-wheels-from-s3 python ./wheels-before # cugraph RAPIDS_PY_WHEEL_NAME="cugraph_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cugraph \ RAPIDS_SHA=${CUGRAPH_NIGHTLY_SHA} \ rapids-download-wheels-from-s3 python ./wheels-before # pylibcugraph RAPIDS_PY_WHEEL_NAME="pylibcugraph_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cugraph \ RAPIDS_SHA=${CUGRAPH_NIGHTLY_SHA} \ rapids-download-wheels-from-s3 python ./wheels-before # cuml RAPIDS_PY_WHEEL_NAME="cuml_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cuml \ RAPIDS_SHA=${CUML_NIGHTLY_SHA} \ rapids-download-wheels-from-s3 python ./wheels-before # --- wheels from CI --- # mkdir -p ./wheels-after export RAPIDS_BUILD_TYPE="pull-request" # libraft RAPIDS_PY_WHEEL_NAME="libraft_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/raft \ RAPIDS_REF_NAME="${RAFT_PR}" \ RAPIDS_SHA="${RAFT_PR_SHA}" \ rapids-download-wheels-from-s3 cpp ./wheels-after # pylibraft RAPIDS_PY_WHEEL_NAME="pylibraft_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/raft \ RAPIDS_REF_NAME="${RAFT_PR}" \ RAPIDS_SHA="${RAFT_PR_SHA}" \ rapids-download-wheels-from-s3 python ./wheels-after # raft-dask RAPIDS_PY_WHEEL_NAME="raft_dask_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/raft \ RAPIDS_REF_NAME="${RAFT_PR}" \ RAPIDS_SHA="${RAFT_PR_SHA}" \ rapids-download-wheels-from-s3 python ./wheels-after # libcugraph RAPIDS_PY_WHEEL_NAME="libcugraph_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cugraph \ RAPIDS_REF_NAME="${CUGRAPH_PR}" \ RAPIDS_SHA="${CUGRAPH_PR_SHA}" \ rapids-download-wheels-from-s3 cpp ./wheels-after # pylibcugraph RAPIDS_PY_WHEEL_NAME="pylibcugraph_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cugraph \ RAPIDS_REF_NAME="${CUGRAPH_PR}" \ RAPIDS_SHA="${CUGRAPH_PR_SHA}" \ rapids-download-wheels-from-s3 python ./wheels-after # cugraph RAPIDS_PY_WHEEL_NAME="cugraph_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cugraph \ RAPIDS_REF_NAME="${CUGRAPH_PR}" \ RAPIDS_SHA="${CUGRAPH_PR_SHA}" \ rapids-download-wheels-from-s3 python ./wheels-after # libcuml RAPIDS_PY_WHEEL_NAME="libcuml_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cuml \ RAPIDS_REF_NAME="${CUML_PR}" \ RAPIDS_SHA="${CUML_PR_SHA}" \ rapids-download-wheels-from-s3 cpp ./wheels-after # cuml RAPIDS_PY_WHEEL_NAME="cuml_${RAPIDS_PY_CUDA_SUFFIX}" \ RAPIDS_REPOSITORY=rapidsai/cuml \ RAPIDS_REF_NAME="${CUML_PR}" \ RAPIDS_SHA="${CUML_PR_SHA}" \ rapids-download-wheels-from-s3 python ./wheels-after pip install pydistcheck pydistcheck \ --inspect \ --select 'distro-too-large-compressed' \ ./wheels-before/*.whl \ | grep -E '^checking|files: | compressed' \ > ./before.txt # get more exact sizes du -sh ./wheels-before/* pydistcheck \ --inspect \ --select 'distro-too-large-compressed' \ ./wheels-after/*.whl \ | grep -E '^checking|files: | compressed' \ > ./after.txt # get more exact sizes du -sh ./wheels-after/* ``` </details> ### How I tested this These other PRs: * rapidsai/devcontainers#435 * rapidsai/cugraph-gnn#110 * rapidsai/cuml#6199 * rapidsai/cugraph#4804

…ml-wheels

jameslamb · 2025-01-17T18:41:04Z

/ok to test

…wheels

jameslamb · 2025-01-17T21:49:08Z

/ok to test

jameslamb · 2025-01-17T22:28:45Z

python/cuml/CMakeLists.txt

+  # libcuml (C++) and cuml (Cython).
+  set(CUML_USE_CUVS_STATIC OFF)
+  set(CUML_EXCLUDE_CUVS_FROM_ALL ON)
+  include(${CUML_CPP_SRC}/cmake/thirdparty/get_cuvs.cmake)


Some of the Cython code has direct uses cuVS:

cuml/python/cuml/cuml/cluster/kmeans_utils.pxd

Line 39 in d95cae5

cdef extern from "cuvs/cluster/kmeans.hpp" namespace \

so cuVS is needed in the build environment for both libcuml and cuml wheels. That means we end up compiling libcuvs.so in every libcuml build AND every cuml build.

Even if with decent cache hit rates, on this PR I've seen that result in it taking on the order of 2.5 hours end-to-end for all the build-libcuml and build-cuml jobs to complete :/

Maybe we need to stop here and try to add a libcuvs wheel?

jameslamb · 2025-01-17T22:37:03Z

ci/validate_wheel.sh

+        )
+    fi
+elif [[ "${package_dir}" == "python/cuml" ]]; then
+    # TODO(jameslamb): why are the CUDA 11 wheels so big???


I haven't found the root cause yet, but the CUDA 11 cuml wheels being produced on this branch are a lot bigger than I'd expect.

Suspect it's related to static linking against the CUDA math wheels, but I'm surprised that the difference could be so large for these Cython extensions.

For context, the Cython extension sizes don't really same to vary much by CUDA version on latest branch-25.02:

CUDA 11.8.0: https://github.com/rapidsai/cuml/actions/runs/12836193902/job/35797302998#step:10:1672

CUDA 12.5.1: https://github.com/rapidsai/cuml/actions/runs/12836193902/job/35797303218#step:10:2754

It's just libcuml++.so driving the big difference in total size on branch-25.02.

CUDA 11.8.0, arm64, Python 3.11

file size * compressed size: 1.1G * uncompressed size: 4.0G * compression space saving: 72.4% contents * directories: 74 * files: 441 (85 compiled) ... largest files * (86.8M) cuml/experimental/fil/fil.cpython-311-aarch64-linux-gnu.so * (86.7M) cuml/fil/fil.cpython-311-aarch64-linux-gnu.so * (85.3M) cuml/ensemble/randomforest_shared.cpython-311-aarch64-linux-gnu.so * (85.2M) cuml/explainer/tree_shap.cpython-311-aarch64-linux-gnu.so * (85.2M) cuml/explainer/kernel_shap.cpython-311-aarch64-linux-gnu.so

(build link)

CUDA 12.5.1, arm64, Python 3.11

file size * compressed size: 8.9M * uncompressed size: 30.1M * compression space saving: 70.4% contents * directories: 74 * files: 441 (85 compiled) ... largest files * (2.6M) cuml/experimental/fil/fil.cpython-311-aarch64-linux-gnu.so * (2.4M) cuml/fil/fil.cpython-311-aarch64-linux-gnu.so * (1.0M) cuml/cluster/hdbscan/hdbscan.cpython-311-aarch64-linux-gnu.so * (0.9M) cuml/svm/linear.cpython-311-aarch64-linux-gnu.so * (0.9M) cuml/manifold/umap.cpython-311-aarch64-linux-gnu.so

(build link)

jameslamb · 2025-01-17T23:05:08Z

python/cuml/CMakeLists.txt

+  # --- RAFT---#
+  # find RAFT before cuVS, to avoid
+  # cuVS CMake defining conflicting versions of targets like 'nvidia::cutlass::cutlass'
+  include(${CUML_CPP_SRC}/cmake/thirdparty/get_raft.cmake)


Without first finding RAFT here, builds fail like this:

-- Found Thrust: /pyenv/versions/3.12.7/lib/python3.12/site-packages/libraft/lib64/rapids/cmake/thrust/thrust-config.cmake (found suitable exact version "2.7.0.0") -- Found CCCL: /pyenv/versions/3.12.7/lib/python3.12/site-packages/libraft/lib64/rapids/cmake/cccl/cccl-config.cmake (found version "2.7.0.0") -- Found nvtx3: /pyenv/versions/3.12.7/lib/python3.12/site-packages/librmm/lib64/cmake/nvtx3/nvtx3-config.cmake (found version "3.1.0") -- Found rmm: /pyenv/versions/3.12.7/lib/python3.12/site-packages/librmm/lib64/cmake/rmm/rmm-config.cmake (found version "25.02.0") CMake Error at /pyenv/versions/3.12.7/lib/python3.12/site-packages/libraft/lib64/cmake/NvidiaCutlass/NvidiaCutlassTargets.cmake:42 (message): Some (but not all) targets in this export set were already defined. Targets Defined: nvidia::cutlass::cutlass Targets not yet defined: nvidia::cutlass::tools::util Call Stack (most recent call first): /pyenv/versions/3.12.7/lib/python3.12/site-packages/libraft/lib64/cmake/NvidiaCutlass/NvidiaCutlassConfig.cmake:9 (include) /pyenv/versions/3.12.7/lib/python3.12/site-packages/cmake/data/share/cmake-3.31/Modules/CMakeFindDependencyMacro.cmake:76 (find_package) /pyenv/versions/3.12.7/lib/python3.12/site-packages/libraft/lib64/cmake/raft/raft-dependencies.cmake:43 (find_dependency) /pyenv/versions/3.12.7/lib/python3.12/site-packages/libraft/lib64/cmake/raft/raft-config.cmake:83 (include) /pyenv/versions/3.12.7/lib/python3.12/site-packages/libcuml/lib64/cmake/cuml/cuml-dependencies.cmake:40 (find_package) /pyenv/versions/3.12.7/lib/python3.12/site-packages/libcuml/lib64/cmake/cuml/cuml-config.cmake:72 (include) CMakeLists.txt:150 (find_package) -- Configuring incomplete, errors occurred! *** CMake configuration failed error: subprocess-exited-with-error

I suspect that's some interaction between RAFT's exports (which do include cutlass)) and cuVS's (code link), but I haven't figured it out yet.

jameslamb added 2 commits December 24, 2024 08:36

initial libcuml wheel

e42f3d1

remove uses of ASSOCIATED_TARGETS, lots of other refactoring

13db0a9

jameslamb added 2 - In Progress Currenty a work in progress 5 - DO NOT MERGE Hold off on merging; see PR for details labels Dec 30, 2024

This comment was marked as resolved.

Sign in to view

github-actions bot added Cython / Python Cython or Python issue CMake CUDA/C++ ci labels Dec 30, 2024

Merge branch 'branch-25.02' into libcuml-wheels

f33e5ed

jameslamb mentioned this pull request Dec 30, 2024

introduce libraft wheels rapidsai/raft#2531

Merged

jameslamb added 4 commits December 31, 2024 08:01

more changes

368faea

Merge branch 'branch-25.02' of github.com:rapidsai/cuml into libcuml-…

bbc4742

…wheels

use static treelite

e99a2b7

merge branch-25.02

b6eeb73

github-actions bot removed the CUDA/C++ label Dec 31, 2024

jameslamb added 3 commits December 31, 2024 13:07

one more merge conflict

8082c45

more fiddling

8434d77

more changes

9bda048

github-actions bot added conda conda issue CUDA/C++ labels Jan 2, 2025

jameslamb added 9 commits January 6, 2025 10:13

more changes

eb14ede

Merge branch 'branch-25.02' into libcuml-wheels

c5d385b

got importing minimally working

8c1f38d

use conda packages from RAFT PR

7c09598

pre-commit

c7ad450

copyright

8d058e6

fix scripts

cbf26a2

fix wheel downloads

dc8a5a4

clone cumlprims_mg in CI

9773c81

use dynamic treelite in conda builds

e35df32

jameslamb added 7 commits January 13, 2025 19:10

refine build dependencies

7853681

merge branch-25.02

7e94d7d

more dependency fixes

ba3387f

fix scripts

1f9ac89

update RAFT commit, build cuml without build isolation

82e3f16

try fixing math wheels linkage for CUDA 11

53c8b35

add use_cuda_wheels

002c0f3

jameslamb mentioned this pull request Jan 14, 2025

Support dynamic linking between RAPIDS wheels rapidsai/build-planning#33

Open

fix build_wheel_cuml.sh again

35c88ac

jameslamb mentioned this pull request Jan 14, 2025

introduce libcugraph wheels rapidsai/cugraph#4804

Merged

jameslamb added 4 commits January 14, 2025 08:43

try to fix build_wheel_cuml.sh again

0e62018

misc. changes

f5b6590

Merge branch 'branch-25.02' of github.com:rapidsai/cuml into libcuml-…

e4c1fad

…wheels

rapids_cuda_init_architectures()

c79617b

bdice and others added 5 commits January 17, 2025 09:54

Revert libraft testing changes.

57db9f7

simplify CMake

e7109c4

Merge branch 'libcuml-wheels' of github.com:jameslamb/cuml into libcu…

e2ae2dc

…ml-wheels

fix typo in rapids_cuda_init_architectures

b17f613

fix SINGLEGPU arg

2623cc9

rapidsai deleted a comment from bdice Jan 17, 2025

jameslamb added 2 commits January 17, 2025 13:47

updates

77c8f14

Merge branch 'branch-25.02' of github.com:rapidsai/cuml into libcuml-…

668e46b

…wheels

jameslamb commented Jan 17, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

WIP: [DO NOT MERGE] introduce libcuml wheels #6199

WIP: [DO NOT MERGE] introduce libcuml wheels #6199

jameslamb commented Dec 30, 2024 •

edited

Loading

This comment was marked as resolved.

jameslamb commented Jan 17, 2025

jameslamb commented Jan 17, 2025

jameslamb Jan 17, 2025

jameslamb Jan 17, 2025

jameslamb Jan 17, 2025

WIP: [DO NOT MERGE] introduce libcuml wheels #6199

Are you sure you want to change the base?

WIP: [DO NOT MERGE] introduce libcuml wheels #6199

Conversation

jameslamb commented Dec 30, 2024 • edited Loading

Notes for Reviewers

Benefits of these changes

Wheel contents

Dependency Flows

Size changes (CUDA 12, Python 3.12, x86_64)

How I tested this

This comment was marked as resolved.

jameslamb commented Jan 17, 2025

jameslamb commented Jan 17, 2025

jameslamb Jan 17, 2025

Choose a reason for hiding this comment

jameslamb Jan 17, 2025

Choose a reason for hiding this comment

jameslamb Jan 17, 2025

Choose a reason for hiding this comment

jameslamb commented Dec 30, 2024 •

edited

Loading