Releases: rapidsai/cugraph
Releases Β· rapidsai/cugraph
v24.04.00
π¨ Breaking Changes
- C API for MNMG ECG (#4194) @naimnv
- Refactor and Fix
test_induced_subgraph_mg
(#4183) @nv-rliu - Removes APIs that have been deprecated and have exceeded the grace period, improves *_warning_wrapper performance (#4176) @rlratzel
- Fixes for pandas 2, latest cudf, and wheel building (#4144) @vyasr
π Bug Fixes
- Update pre-commit-hooks to v0.0.3 (#4264) @KyleFromNVIDIA
- Removes (unintentional?) runtime dependency on
packaging
from nx-cugraph (#4240) @rlratzel - Fix kv_store_t build error when cuCollection is used as a backend and the value type is thrust::tuple (#4238) @seunghwak
- MTMG gather use of streams (#4237) @ChuckHastings
- remove NVTX calls in libcugraph_etl (#4229) @ChuckHastings
- Update cuGraph-PyG GraphSAGE Examples (#4224) @alexbarghi-nv
- transform_e bug fix in edge masking (#4221) @seunghwak
- Remove Usage of Experimental Datasets Package (#4220) @nv-rliu
- Fix pos_list api for force atlas (#4219) @hlinsen
- Fix
louvain_communities
,PropertyGraph
, cudfcolumn.full
, dgl 2.1.0 CI failures (#4215) @eriknw - Remove CUGRAPH_EXPECTS(!(this->has_edge_mask()), unimplemented.) (#4214) @naimnv
- fix scatter to handle empty graphs (#4193) @tingyu66
- Refactor and Fix
test_induced_subgraph_mg
(#4183) @nv-rliu - Update SG notebook (#4169) @naimnv
- Use input rng_state rather than creating a local one (#4168) @naimnv
- Fix
FutureWarnings
in Graph Creation (#4167) @nv-rliu - Fixes for pandas 2, latest cudf, and wheel building (#4144) @vyasr
- handle more RAPIDS version formats in update-version.sh, refactor dependencies.yaml (#4109) @jameslamb
- Mark kernels as internal (#4098) @robertmaynard
π Documentation
- Add Python 3.11 to Conda docs (#4213) @jakirkham
- Update installation docs for CUDA 12. (#4207) @bdice
- Fix in Docs to address question asked in issue 4182 (#4204) @acostadon
- [docs] fix GNN typo (#4196) @aMahanna
- Update README for cugraph notebooks (#4180) @naimnv
- Starting work on blog links and nx_cugraph docs (#4160) @acostadon
π New Features
- Graph coloring and MIS (#4211) @naimnv
- Enable edge masking in the remaining primitives (#4186) @seunghwak
- Update per_v_random_select_transform_outgoing_e to support edge masking (#4174) @seunghwak
- cugraph examples (#4128) @naimnv
- Enable edge masking in additional primitives (#4126) @seunghwak
- Support CUDA 12.2 (#4088) @jameslamb
- cuGraph-DGL and WholeGraph Performance Testing with Feature Store Performance Improvements (#4081) @alexbarghi-nv
π οΈ Improvements
- Use
conda env create --yes
instead of--force
(#4286) @bdice - Always scatter in FP32 in TPConv (#4260) @tingyu66
- Remove outdated conda_build.sh. (#4257) @bdice
- update example build script with targets and options/help (#4254) @naimnv
- Refactor test utility functions and update SG & MG core_number tests to test with edge masking (#4253) @seunghwak
- nx-cugraph: add
is_negatively_weighted
(#4242) @eriknw - Add upper bound to prevent usage of NumPy 2 (#4233) @bdice
- Adds nx-cugraph benchmarks for APIs added to prior releases that were never benchmarked (#4228) @rlratzel
- Make external vertex and edge shuffling function public (#4227) @naimnv
- Updates nx-cugraph
README
for latest h/w, CUDA, python, NX requirements, moves updater to pre-commit (#4225) @rlratzel - Enable pytest failures on warnings from cudf (#4223) @mroeschke
- Added a weighted example to the jaccard notebook (#4222) @acostadon
- Use public pandas and cudf APIs where possible (#4218) @mroeschke
- Remove hard-coding of RAPIDS version where possible (#4217) @KyleFromNVIDIA
- Treat nx-cugraph, cugraph-pyg, cugraph-dgl, and cugraph-equivariant CI artifacts as pure wheels (#4216) @bdice
- Add degrees to C API (#4212) @ChuckHastings
- Add additional
kwargs
to GATConv (#4210) @tingyu66 - Add
DeprecationWarning
tobatched_ego_graphs
(#4209) @nv-rliu - Trim trailing whitespace (#4206) @bdice
- Automate C++ include file grouping and ordering using clang-format (#4205) @harrism
- nx-cugraph: add more shortest path algorithms (#4199) @eriknw
- Refactor MG Centrality Tests (#4197) @nv-rliu
- Add support for Python 3.11, require NumPy 1.23+ (#4195) @jameslamb
- C API for MNMG ECG (#4194) @naimnv
- nx-cugraph: update
get_info
function keys to NetworkX 3.3 (#4192) @eriknw - Adds additional detail to Jaccard notebook (#4189) @rlratzel
- Update devcontainers to CUDA Toolkit 12.2 (#4184) @trxcllnt
- Update
louvain_communities
to match NetworkX 3.3 (added max_level) (#4177) @eriknw - Removes APIs that have been deprecated and have exceeded the grace period, improves *_warning_wrapper performance (#4176) @rlratzel
- target branch-24.04 for GitHub Actions workflows (#4175) @jameslamb
- Add cuda-nvtx-dev, add missing CUDA library dependencies. (#4162) @bdice
- Enforce matching type (#4161) @jnke2016
- Add new all-pairs similarity algorithm (#4158) @ChuckHastings
- nx-cugraph: automatically generate trees in README.md (#4156) @eriknw
- Add environment-agnostic scripts for running ctests and pytests (#4152) @trxcllnt
- Ensure that
ctest
is called with--no-tests=error
. (#4148) @bdice - Update ops-bot.yaml (#4146) @AyodeAwe
- Replace local copyright check with pre-commit-hooks verify-copyright (#4130) @KyleFromNVIDIA
- Re-run
ci/release/update-version.sh 24.04.00
(#4127) @jakirkham - Address primitives performance regression with the added edge masking support (when edge masking is not in use) (#4119) @seunghwak
- Update cugraph for compatibility with the latest cuco (#4111) @PointKernel
- Add a new notebook for SNMG benchmark runs (#4091) @naimnv
- Ktruss implementation (#4059) @jnke2016
[NIGHTLY] v24.06.00
π Links
π¨ Breaking Changes
- Address dask_cudf.read_csv chunksize deprecation (#4379) @mroeschke
- Convert device_memory_resource* to device_async_resource_ref (#4365) @harrism
π Bug Fixes
- Address dask_cudf.read_csv chunksize deprecation (#4379) @mroeschke
- Fix bad Leiden result (#4376) @ChuckHastings
- Erdos-Renyi generator had bad logic in thrust calls (#4362) @ChuckHastings
- Change mamba option --force to --yes (#4357) @naimnv
- Fix false C++ HITS test failure on H100 (#4352) @seunghwak
- Bug fix in cugraph::detail::extract_transform_v_froniter_e (#4320) @seunghwak
- Refactor CMakeLists.txt to better express usage requirements (#4309) @robertmaynard
- [BUG] Always Persist Dask DataFrames in cuGraph-DGL Graph Storage (#4296) @alexbarghi-nv
- Fix Bug in MG
ego_graph
(#4262) @nv-rliu - [FIX] Remove Unnecessary CUDA Check Causing Deprecation Warning (#4239) @alexbarghi-nv
π Documentation
- Removed obsolete methods from rst file (#4331) @acostadon
- Updates the nx-cugraph README.md with the latest upstream URLs (#4321) @rlratzel
- Fix Table for Synthetic Data notebook (#4311) @nv-rliu
- Fix Docstring for
from_dask_cudf_edgelist
(#4310) @nv-rliu - fix PyPI URL in docs (#4282) @jameslamb
- Notebook fixes to formulas and outputs (#4272) @acostadon
π New Features
- Always use a static gtest (#4344) @robertmaynard
- Update GNN sampling post processing functions to optionally consider seed vertices in renumbering (#4329) @seunghwak
- [FEA] cuGraph GNN NCCL-only Setup and Distributed Sampling (#4278) @alexbarghi-nv
π οΈ Improvements
- Update
k_truss
to work with many connected components (#4383) @eriknw - Migrate to
{{ stdlib("c") }}
(#4374) @hcho3 - Convert device_memory_resource* to device_async_resource_ref (#4365) @harrism
- Add missing
<optional>
include (#4353) @trxcllnt nx-cugraph
: supportshould_run
that was added in NetworkX 3.3 (#4348) @eriknw- Update generation and flattening of dendrogram in Leiden (#4347) @ChuckHastings
- Update pre-commit hooks to include all C and C++ files (#4332) @ChuckHastings
- Use appropriate edge src and dst properties based on is_multi_gpu flag (#4319) @naimnv
- Test GNN Examples and Add CUDA 12 Testing (#4317) @alexbarghi-nv
- Adds missing files to
update-version.sh
(#4314) @AyodeAwe - Get rid of sentinel namespace (#4274) @PointKernel
- Update more C++ algorithm tests to include edge masked input graphs. (#4267) @seunghwak
- Update triangle count to internally use edge masking and update Triangle counting & K-hop neighbor tests to include edge masked graph view inputs (#4261) @seunghwak
- Add Additional Checks to
get_edgelist
andget_dask_edgelist
(#4256) @nv-rliu
v24.02.00
π¨ Breaking Changes
- Remove Experimental Wrappers from GNN Code (#4070) @alexbarghi-nv
- Switch to scikit-build-core (#4053) @vyasr
- Update to CCCL 2.2.0. (#4052) @bdice
π Bug Fixes
- Revert "Exclude tests from builds (#4147)" (#4157) @raydouglass
- Exclude tests from builds (#4147) @vyasr
- Constraint pytorch-dependent wheel test to only run on amd64 (#4133) @tingyu66
- Removes the
networkx_algorithm
decorator to all SCC functions to disable dispatching to them (#4120) @rlratzel - Correct
cugraph-pyg
package name used in wheels and fix test script (#4083) @tingyu66 - Fix Jaccard hang (#4080) @jnke2016
- Fix OOB error, BFS C API should validate that the source vertex is a valid vertex (#4077) @ChuckHastings
- [BUG]Fix non-type template parameter to cugraph::relabel (#4064) @naimnv
- Fix MG weighted similarity test failure (#4054) @seunghwak
- MG C-API test failure fixes (#4047) @seunghwak
- Add a barrier before cugraph Graph creation (#4046) @VibhuJawa
- Fix % 0 bug in MG_SELECT_RANDOM_VERTICES test (#4034) @seunghwak
- Branch 24.02 merge branch 23.12 (#4012) @vyasr
π Documentation
- Updates nx-cugraph README.md with latest algos (#4135) @rlratzel
- corrected links in C API and added groups for support functions (#4131) @acostadon
- Forward-merge branch-23.12 to branch-24.02 (#4049) @GPUtester
π New Features
- Implement has_edge() & compute_multiplicity() (#4096) @seunghwak
- Update per_v_transform_reduce_incoming|outgoing_e to support edge masking (#4085) @seunghwak
- Remove Experimental Wrappers from GNN Code (#4070) @alexbarghi-nv
- MNMG ECG (#4030) @naimnv
- Replace graph_view.hpp::number_of_edges with compute_number_of_edges (#4026) @seunghwak
- Update count_if_e, transform_reduce_e, and transform_e to support edge masking (#4001) @seunghwak
- Sampling Performance Testing (#3584) @alexbarghi-nv
π οΈ Improvements
- Adds option to rapids_cpm_find for raft to disable hnswlib feature, adds updates for pytest 8 compat, temporarily skips IO intensive test in CI (#4121) @rlratzel
- Adds benchmarks for additional nx-cugraph 24.02 algos (#4112) @rlratzel
- nx-cugraph: use coverage to ensure all algorithms were run (#4108) @eriknw
- nx-cugraph: rename
plc=
to_plc=
(#4106) @eriknw - nx-cugraph: add
complement
andreverse
(#4103) @eriknw - nx-cugraph: add
core_number
(undirected graphs only) (#4100) @eriknw - nx-cugraph: add
is_tree
, etc. (#4097) @eriknw - Optimize the drop-duplicate functionality (#4095) @jnke2016
- nx-cugraph: add triangles and clustering algorithms (#4093) @eriknw
- nx-cugraph: PLC now handles isolated nodes; clean up our workarounds (#4092) @eriknw
- Remove usages of rapids-env-update (#4090) @KyleFromNVIDIA
- Provide explicit pool sizes and avoid RMM detail APIs (#4086) @harrism
- refactor CUDA versions in dependencies.yaml (#4084) @jameslamb
- build wheels for
cugraph-dgl
andcugraph-pyg
(#4075) @tingyu66 - Match weight-sharing option of GATConv in DGL (#4074) @tingyu66
- nx-cugraph: add weakly connected components (#4071) @eriknw
- nx-cugraph: indicate which plc algorithms are used and version_added (#4069) @eriknw
- Adds
nx-cugraph
benchmarks for 23.12 algos (SSSP, pagerank, hits, katz_centrality, degree_centrality, eigenvector_centrality) (#4065) @rlratzel nx-cugraph
: addto_undirected
method; add reciprocity algorithms (#4063) @eriknw- Switch to scikit-build-core (#4053) @vyasr
- Update to CCCL 2.2.0. (#4052) @bdice
- Prevent
actions/labeler
from addingLabel Checker
labels (#4048) @ajschmidt8 - Update dependencies.yaml to new pip index (#4045) @vyasr
- Remove checks for Pascal, no longer supported (#4044) @ChuckHastings
- Fix HITS convergence error. (#4043) @seunghwak
- Test select_random_vertices for all possible values of flags (#4042) @naimnv
- Remove CUGRAPH_BUILD_WHEELS and standardize Python builds (#4041) @vyasr
- Create
cugraph-equivariant
package (#4036) @tingyu66 - [FEA]: Add DASK edgelist and graph support to the Dataset API (#4035) @huiyuxie
- Add support for Louvain to MTMG (#4033) @ChuckHastings
- Clean up self-loop and multi-edge removal logic (#4032) @ChuckHastings
- Mtmg updates for rmm (#4031) @ChuckHastings
- nx-cugraph: adds
ancestors
,descendants
, and BFS algos (#4029) @eriknw - nx-cugraph: update usage of
nodes_or_number
for nx compat (#4028) @eriknw - Removes unsupported
setup.py
calls, cleans up text (#4024) @rlratzel - Resolves conflicts from forward-merging branch-23.12 into branch-24.02 (#4020) @rlratzel
- Add
HeteroGATConv
tocugraph-pyg
(#3914) @tingyu66 - Update for CCCL 2.x (#3862) @seunghwak
v23.12.00
π¨ Breaking Changes
- [BUG] Restore the original default order of CSR, which does not reverse edges in cuGraph-PyG (#3980) @alexbarghi-nv
Resultset
andDataset
Refactors (#3957) @nv-rliu- Moves more MG graph ETL to libcugraph and re-enables MG tests in CI (#3941) @jnke2016
π Bug Fixes
- Pin actions/labeler to v4 (#4038) @raydouglass
- Find rmm before cuco (#4011) @vyasr
- Pin to minor versions of packages outside the cuGraph repository. (#4004) @bdice
- Move MTMG_TEST to MG tests block (#3993) @naimnv
- Fix Leiden refinement phase (#3990) @naimnv
- [BUG] Fix Graph Construction From Pandas in cuGraph-PyG (#3985) @alexbarghi-nv
- [BUG] Restore the original default order of CSR, which does not reverse edges in cuGraph-PyG (#3980) @alexbarghi-nv
- Fix eigenvector testing and HITS testing discrepancies (#3979) @ChuckHastings
- [BUG] Fix Incorrect Edge Index, Directory Selection in cuGraph-PyG Loader (#3978) @alexbarghi-nv
- [BUG] Check if Dask has quit to avoid throwing an exception and triggering a segfault on ddp exit (#3961) @alexbarghi-nv
- nx-cugraph: xfail test_louvain.py:test_threshold in Python 3.9 (#3944) @eriknw
π Documentation
- [DOC]: Fix invalid links and add materials to notebooks (#4002) @huiyuxie
- Update Broken Links in README.md (#3924) @nv-rliu
π New Features
- Implement the transform_e primitive (to update property values for all edges) (#3917) @seunghwak
- Update the neighbor intersection primitive to support edge masking. (#3550) @seunghwak
π οΈ Improvements
- Correct defect found in DLFW testing (#4021) @ChuckHastings
nx-cugraph
README update: adds missingconnected_components
algo to table (#4019) @rlratzel- Build concurrency for nightly and merge triggers (#4009) @bdice
- Support
drop_last
Argument in cuGraph-PyG Loader (#3995) @alexbarghi-nv - Adds
update-version.sh
support for recently added files containing RAPIDS versions (#3994) @rlratzel - Use new
rapids-dask-dependency
metapackage for managingdask
versions (#3991) @galipremsagar - Fixes to nx-cugraph README: fixes typos, updates link to NX backend docs (#3989) @rlratzel
- Address FIXMEs (#3988) @seunghwak
- Updates README file to include nx-cugraph user documentation, adds nx-cugraph to main README (#3984) @rlratzel
- Update C API graph creation function signatures (#3982) @ChuckHastings
- [REVIEW]Optimize cugraph-DGL csc codepath (#3977) @VibhuJawa
- nx-cugraph: add SSSP (unweighted) (#3976) @eriknw
- CuGraph compatibility fixes (#3973) @brandon-b-miller
- Skip certain
cugraph-pyg
tests when torch-sparse is not available (#3970) @tingyu66 - nx-cugraph: add
eigenvector_centrality
,katz_centrality
,hits
,pagerank
(#3968) @eriknw - Cut peak memory footprint in graph creation (#3966) @seunghwak
- nx-cugraph: add CC for undirected graphs to fix k-truss (#3965) @eriknw
- Skip certain
cugraph-pyg
tests whentorch_sparse
is not available (#3962) @tingyu66 Resultset
andDataset
Refactors (#3957) @nv-rliu- Download
xml
docs artifact through CloudFront endpoint (#3955) @AyodeAwe - Add many graph generators to nx-cugraph (#3954) @eriknw
- Unpin
dask
anddistributed
for23.12
development (#3953) @galipremsagar - Errors compiling for DLFW on CUDA 12.3 (#3952) @ChuckHastings
- nx-cugraph: add k_truss and degree centralities (#3945) @eriknw
- nx-cugraph: handle seed argument in edge_betweenness_centrality (#3943) @eriknw
- Moves more MG graph ETL to libcugraph and re-enables MG tests in CI (#3941) @jnke2016
- Temporarily disable mg testing (#3940) @jnke2016
- adding C/C++ API docs (#3938) @BradReesWork
- Add multigraph support to nx-cugraph (#3934) @eriknw
- Setup Consistent Nightly Versions for Pip and Conda (#3933) @divyegala
- MTMG multi node (#3932) @ChuckHastings
- Use branch-23.12 workflows. (#3928) @bdice
- Fix an issue occurring in the cuGraph-DGL example for "mixed" mode. (#3927) @drivanov
- Updating Docs (#3923) @BradReesWork
- Forward-merge branch-23.10 to branch-23.12 (#3919) @nv-rliu
- new build all option (#3916) @BradReesWork
- Silence spurious compiler warnings (#3913) @seunghwak
- Link wholegrah and cugraphops XML docs (#3906) @AyodeAwe
- Updates to 23.12 (#3905) @raydouglass
- Forward-merge branch-23.10 to branch-23.12 (#3904) @GPUtester
- Build CUDA 12.0 ARM conda packages. (#3903) @bdice
- Merge branch-23.10 into branch-23.12 (#3898) @rlratzel
- Some MTMG code cleanup and small optimizations (#3894) @ChuckHastings
- Enable parallel mode (#3875) @jnke2016
- Adds benchmarks for
nx-cugraph
(#3854) @rlratzel - Add nx-cugraph notebook for showing accelerated networkX APIs (#3830) @betochimas
v23.10.00
π¨ Breaking Changes
- Rename
cugraph-nx
tonx-cugraph
(#3840) @eriknw - Remove legacy betweenness centrality (#3829) @jnke2016
- Remove Deprecated Sampling Options (#3816) @alexbarghi-nv
- cuGraph-PyG Loader Improvements (#3795) @alexbarghi-nv
- Expose threshold in louvain (#3792) @ChuckHastings
- Fix ValueError Caused By Batches With No Samples (#3789) @alexbarghi-nv
- Update to Cython 3.0.0 (#3716) @vyasr
π Bug Fixes
- Add wget to test_notebook dependencies (#3918) @raydouglass
- Increase dask-related timeouts for CI testing (#3907) @jnke2016
- Remove
dask_cudf
dataframe for the_make_plc_graph
while creatingcugraph.Graph
(#3895) @VibhuJawa - Adds logic to handle isolated vertices at python layer (#3886) @naimnv
- Update Allocator Selection in cuGraph-DGL Example (#3877) @alexbarghi-nv
- Add file to update-version.sh (#3870) @raydouglass
- Fix torch seed in
cugraph-dgl
and-pyg
tests for conv layers (#3869) @tingyu66 - MFG C++ code bug fix (#3865) @seunghwak
- Fix subtle memory leak in nbr_intersection primitive (#3858) @ChuckHastings
- Uses
conda mambabuild
rather thanmamba mambabuild
(#3853) @rlratzel - Remove the assumption made on the client data's keys (#3835) @jnke2016
- Disable mg tests (#3833) @naimnv
- Refactor python code for similarity algos to use latest CAPI (#3828) @naimnv
- [BUG] Fix Batch Renumbering of Empty Batches (#3823) @alexbarghi-nv
- Temporarily disable the deletion of the dask dataframe (#3814) @jnke2016
- Fix OD shortest distance matrix computation test failures. (#3813) @seunghwak
- Use rapidsai/ci:cuda11.8.0-ubuntu22.04-py3.10 for docs build (#3811) @naimnv
- Fix ValueError Caused By Batches With No Samples (#3789) @alexbarghi-nv
- Update
python_run_cugraph
independencies.yaml
(#3781) @nv-rliu - Fixes
KeyError
forget_two_hop_neighbors
when called with a small start vertices list (#3778) @rlratzel
π Documentation
π New Features
- WholeGraph Feature Store for cuGraph-PyG and cuGraph-DGL (#3874) @alexbarghi-nv
- similarity notebook to compare link prediction algos (#3868) @acostadon
- adding dining preference dataset (#3866) @acostadon
- Integrate C++ Renumbering and Compression (#3841) @alexbarghi-nv
- Sampling post processing functions to accelerate MFG creation. (#3815) @seunghwak
- [REVIEW] Add Pure DGL Dataloading benchmark (#3660) @VibhuJawa
π οΈ Improvements
- nx-cugraph: handle louvain with isolated nodes (#3897) @eriknw
- Pin
dask
anddistributed
for23.10
release (#3896) @galipremsagar - Updates the source build docs to include libcugraphops as a build prerequisite (#3893) @rlratzel
- fixes force atlas to allow string as vertex names (#3891) @acostadon
- Integrate renumbering and compression to
cugraph-dgl
to accelerate MFG creation (#3887) @tingyu66 - Enable weights for MG similarity algorithms (#3879) @jnke2016
- cuGraph-PyG MFG Creation and Conversion (#3873) @alexbarghi-nv
- Update image names (#3867) @AyodeAwe
- Update to clang 16.0.6. (#3859) @bdice
- Updates to build and test
nx-cugraph
wheel as part of CI and nightly workflows (#3852) @rlratzel - Update
cugraph-dgl
conv layers to use improved graph class (#3849) @tingyu66 - Add entry point to tell NetworkX about nx-cugraph without importing it. (#3848) @eriknw
- [IMP] Add ability to get batch size from the loader in cuGraph-PyG (#3846) @alexbarghi-nv
- Refactor legacy k truss (#3843) @jnke2016
- Use new
raft::compiled_static
targets (#3842) @divyegala - Rename
cugraph-nx
tonx-cugraph
(#3840) @eriknw - Add cuGraph devcontainers (#3838) @trxcllnt
- Enable temporarily disabled MG tests (#3837) @naimnv
- Remove legacy betweenness centrality (#3829) @jnke2016
- Use
copy-pr-bot
(#3827) @ajschmidt8 - Update README.md (#3826) @lmeyerov
- Adding metadata getter methods to datasets API (#3821) @nv-rliu
- Unpin
dask
anddistributed
for23.10
development (#3818) @galipremsagar - Remove Deprecated Sampling Options (#3816) @alexbarghi-nv
- [REVIEW] Cugraph dgl block improvements (#3810) @VibhuJawa
- Simplify wheel build scripts and allow alphas of RAPIDS dependencies (#3809) @vyasr
- Allow cugraph-nx to run networkx tests for nx versions 3.0, 3.1, and 3.2 (#3808) @eriknw
- Add
louvain_communities
to cugraph-nx (#3803) @eriknw - Adds missing copyright and license text to init.py package files (#3799) @rlratzel
- cuGraph-PyG Loader Improvements (#3795) @alexbarghi-nv
- Adds updates to build wheel and conda packages for
cugraph-nx
(#3793) @rlratzel - Expose threshold in louvain (#3792) @ChuckHastings
- Allow models to use a lightweight sparse structure (#3782) @tingyu66
- Clean-up old testing conventions in
test_ecg.py
(#3779) @nv-rliu - Calling
dataset.get_edgelist()
returns a copy of an edge list instead of global (#3777) @nv-rliu - Update dgl benchmarks (#3775) @VibhuJawa
- Forward-merge branch-23.08 to branch-23.10 (#3774) @nv-rliu
- Migrate upstream models to
cugraph-pyg
(#3763) @tingyu66 - Branch 23.10 merge 23.08 (#3743) @vyasr
- Update to Cython 3.0.0 (#3716) @vyasr
- Testing util improvements and refactoring (#3705) @betochimas
- Add new cugraph-nx package (networkx backend using pylibcugraph) (#3614) @eriknw
- New mtmg API for integration (#3521) @ChuckHastings
v23.08.00
π¨ Breaking Changes
- Change the renumber_sampled_edgelist function behavior. (#3762) @seunghwak
- PLC and Python Support for Sample-Side MFG Creation (#3734) @alexbarghi-nv
- Stop using setup.py in build.sh (#3704) @vyasr
- Refactor edge betweenness centrality (#3672) @jnke2016
- [FIX] Fix the hang in cuGraph Python Uniform Neighbor Sample, Add Logging to Bulk Sampler (#3669) @alexbarghi-nv
π Bug Fixes
- Change the renumber_sampled_edgelist function behavior. (#3762) @seunghwak
- Fix bug discovered in Jaccard testing (#3758) @ChuckHastings
- fix inconsistent graph properties between the SG and the MG API (#3757) @jnke2016
- Fixes options for
--pydevelop
to remove unneeded CWD path ("."), restores use ofsetup.py
temporarily for develop builds (#3747) @rlratzel - Fix sampling call parameters if compiled with -DNO_CUGRAPH_OPS (#3729) @ChuckHastings
- Fix primitive bug discovered in MG edge betweenness centrality testing (#3723) @ChuckHastings
- Reorder dependencies.yaml channels (#3721) @raydouglass
- [BUG] Fix namesapce to default_hash and hash_functions (#3711) @naimnv
- [BUG] Fix Bulk Sampling Test Issue (#3701) @alexbarghi-nv
- Make
pylibcugraphops
optional imports incugraph-dgl
and-pyg
(#3693) @tingyu66 - [FIX] Rename
cugraph-ops
symbols (refactoring) and update GHA workflows to call pytest viapython -m pytest
(#3688) @naimnv - [FIX] Fix the hang in cuGraph Python Uniform Neighbor Sample, Add Logging to Bulk Sampler (#3669) @alexbarghi-nv
- force atlas notebook changes to run in cugraph 23.08 container. (#3656) @acostadon
π Documentation
- this fixes github links in cugraph, cugraph-dgl and cugraph-pyg (#3650) @acostadon
- Fix minor typo in README.md (#3636) @akasper
- Created landing spot for centrality and similarity algorithms (#3620) @acostadon
π New Features
- Compute shortest distances between given sets of origins and destinations for large diameter graphs (#3741) @seunghwak
- Update primitive to compute weighted Jaccard, Sorensen and Overlap similarity (#3728) @naimnv
- Add CUDA 12.0 conda environment. (#3725) @bdice
- Renumber utility function for sampling output (#3707) @seunghwak
- Integrate C++ Sampling Source Behavior Updates (#3699) @alexbarghi-nv
- Adds
fail_on_nonconvergence
option topagerank
to provide pagerank results even on non-convergence (#3639) @rlratzel - Add Benchmark for Bulk Sampling (#3628) @alexbarghi-nv
- cugraph: Build CUDA 12 packages (#3456) @vyasr
π οΈ Improvements
- Pin
dask
anddistributed
for23.08
release (#3761) @galipremsagar - Fix
build.yaml
workflow (#3756) @ajschmidt8 - Support MFG creation on sampling gpus for cugraph dgl (#3742) @VibhuJawa
- PLC and Python Support for Sample-Side MFG Creation (#3734) @alexbarghi-nv
- Switch to new wheel building pipeline (#3731) @vyasr
- Remove RAFT specialization. (#3727) @bdice
- C API for renumbering the samples (#3724) @ChuckHastings
- Only run cugraph conda CI for CUDA 11. (#3713) @bdice
- Promote
Datasets
to stable and clean-up unit tests (#3712) @nv-rliu - [BUG] Unsupported graph for similiarity algos (#3710) @jnke2016
- Stop using setup.py in build.sh (#3704) @vyasr
- [WIP] Make edge ids optional (#3702) @VibhuJawa
- Use rapids-cmake testing to run tests in parallel (#3697) @robertmaynard
- Sampling modifications to support PyG and DGL options (#3696) @ChuckHastings
- Include cuCollection public header for hash functions (#3694) @seunghwak
- Refactor edge betweenness centrality (#3672) @jnke2016
- Refactor RMAT (#3662) @jnke2016
- [REVIEW] Optimize bulk sampling (#3661) @VibhuJawa
- Update to CMake 3.26.4 (#3648) @vyasr
- Optimize cugraph-dgl MFG creation (#3646) @VibhuJawa
- use rapids-upload-docs script (#3640) @AyodeAwe
- Fix dependency versions for
23.08
(#3638) @ajschmidt8 - Unpin
dask
anddistributed
for development (#3634) @galipremsagar - Remove documentation build scripts for Jenkins (#3627) @ajschmidt8
- Unpin scikit-build upper bound (#3609) @vyasr
- Implement C++ Edge Betweenness Centrality (#3602) @ChuckHastings
v23.06.02
π¨ Breaking Changes
- [BUG] Fix Incorrect File Selection in cuGraph-PyG Loader (#3599) @alexbarghi-nv
- Remove legacy leiden (#3581) @ChuckHastings
- [IMP] Match Default PyG Hop ID Behavior in cuGraph-PyG (#3565) @alexbarghi-nv
- [IMP] Sample with Offsets in the Bulk Sampler (#3524) @alexbarghi-nv
- Dropping Python 3.8 (#3505) @divyegala
- Remove legacy renumber and shuffle calls from cython.cu (#3467) @ChuckHastings
- Remove legacy implementation of induce subgraph (#3464) @ChuckHastings
π Bug Fixes
- Fix broadcast root during the replication call (#3655) @jnke2016
- [HOTFIX] Fix random walks error identified in CUDA-12 testing (#3647) @ChuckHastings
- Fix MG Test Failing due to Removal of np.float (#3621) @alexbarghi-nv
- fix logic for shuffling results (#3619) @ChuckHastings
- [BUG] Fix Calls to cudf.DataFrame/Series.unique that relied on old behavior (#3616) @alexbarghi-nv
- correct dgl version in
cugraph-dgl
conda recipe (#3612) @tingyu66 - [BUG] Fix Issue in cuGraph-PyG Tests Blocking CI (#3607) @alexbarghi-nv
- [BUG] Critical: Fix cuGraph-PyG Edge Index Renumbering for Single-Edge Graphs (#3605) @alexbarghi-nv
- [BUG] Skip Empty Partitions in Bulk Sample Writing (#3600) @alexbarghi-nv
- [BUG] Fix Incorrect File Selection in cuGraph-PyG Loader (#3599) @alexbarghi-nv
- Fix SSSP bug (#3597) @jnke2016
- update cudf column constructor calls (#3592) @ChuckHastings
- Fix one more path to cugraphops in build workflow (#3554) @vyasr
- Fix path to cugraphops in build workflow (#3547) @vyasr
- Update dgl APIs for v1.1.0 (#3546) @tingyu66
- Pin to scikit-build<17.2 (#3538) @vyasr
- Correct results from sampling when grouping batches on specific GPUs (#3517) @ChuckHastings
- [FIX] Match the PyG API for Node Input to the Loader (#3514) @alexbarghi-nv
- Correct MG Leiden and SCC tests (#3509) @ChuckHastings
- per_v_transform_reduce_incoming|outgoing_e bug fix (when we're using (key, value) pairs to store edge src|dst property values) (#3508) @seunghwak
- Updates to allow python benchmarks to run on additional datasets by default (#3506) @rlratzel
- [BUG] Fix Intermittent Error when Converting cuDF DataFrame to Tensor by Converting to cuPy Array First (#3498) @alexbarghi-nv
- [FIX] Update cugraph-PyG Dependencies to include cuGraph (#3497) @alexbarghi-nv
- Fix graph_properties_t's members order (#3484) @naimnv
- Fix issue with latest rapids-make (#3481) @ChuckHastings
- Branch 23.06 Fix Forward Merge (#3462) @alexbarghi-nv
- Update raft dependency to 23.06 (#3410) @ChuckHastings
π Documentation
- updated cugraph Demo notebooks for 23.06 (#3558) @acostadon
- cugraph-ops license (#3553) @BradReesWork
- Notebook clean-up and run verification (#3551) @acostadon
- Updates contributing steps to add copyright and license text inclusion instruction (#3519) @rlratzel
- Fixed notebook links in algorithm and cugraph notebook pages (#3515) @acostadon
- adding cugraph-ops (#3488) @BradReesWork
- Sphinx updates (#3468) @BradReesWork
π New Features
- [REVIEW] Add MNMG with training (#3603) @VibhuJawa
- MG Leiden and MG MIS (#3582) @naimnv
- graph primitive transform_e (#3548) @seunghwak
- Support CUDA 12.0 for pip wheels (#3544) @divyegala
- Updates pytest benchmarks to use synthetic data and multi-GPUs (#3540) @rlratzel
- Enable edge masking (#3522) @seunghwak
- [REVIEW] Profile graph creation runtime and memory footprint (#3518) @VibhuJawa
- Bipartite R-mat graph generation. (#3512) @seunghwak
- Dropping Python 3.8 (#3505) @divyegala
- Creates Notebook that runs Multi-GPU versions of Jaccard, Sorenson and overlap. (#3504) @acostadon
- [cugraph-dgl] Add support for bipartite node features and optional edge features in GATConv (#3503) @tingyu66
- [cugraph-dgl] Add TransformerConv (#3501) @tingyu66
- [cugraph-pyg] Add TransformerConv and support for bipartite node features in GATConv (#3489) @tingyu66
- Branch 23.06 resolve merge conflict for forward merge (#3409) @alexbarghi-nv
- Refactor Leiden (#3327) @jnke2016
π οΈ Improvements
- Disables memory usage assertions (#3641) @rlratzel
- Refresh requirements (#3622) @jakirkham
- Pr3266 continue (optional arg for weight attribute for Nx graphs in
sssp
) (#3611) @eriknw - Enables MG python tests using a single-GPU LocalCUDACluster in CI (#3596) @rlratzel
- UVM notebook update and add tracker for notebooks to readme (#3595) @acostadon
- [REVIEW] Skip adding edge types, edge weights (#3583) @VibhuJawa
- Remove legacy leiden (#3581) @ChuckHastings
- run docs nightly too (#3568) @AyodeAwe
- include hop as part of the sort criteria for sampling results (#3567) @ChuckHastings
- Add MG python implementation of Leiden (#3566) @jnke2016
- [IMP] Match Default PyG Hop ID Behavior in cuGraph-PyG (#3565) @alexbarghi-nv
- Switch back to using primary shared-action-workflows branch (#3562) @vyasr
- removed deprecated calls and modified demo notebooks to run with 23.06 (#3561) @acostadon
- add unit test for checking is_symmetric is valid, update documentatio⦠(#3559) @ChuckHastings
- Update recipes to GTest version >=1.13.0 (#3549) @bdice
- Improve memory footprint and performance of graph creation (#3542) @VibhuJawa
- Update cupy dependency (#3539) @vyasr
- Perform expensive edge list check in create_graph_from_edgelist() (#3533) @seunghwak
- Enable sccache hits from local builds (#3526) @AyodeAwe
- Build wheels using new single image workflow (#3525) @vyasr
- [IMP] Sample with Offsets in the Bulk Sampler (#3524) @alexbarghi-nv
- Revert shared-action-workflows pin (#3523) @divyegala
- [FIX] fix cugraphops namespace (#3520) @stadlmax
- Add support in C API for handling unweighted graphs in algorithms that expect weights (#3513) @ChuckHastings
- Changes to support gtest version 1.11 (#3511) @ChuckHastings
- update docs (#3510) @BradReesWork
- Remove usage of rapids-get-rapids-version-from-git (#3502) @jjacobelli
- Remove Dummy Edge Weights, Support Specifying Edge Ids/Edge Types/Weights Separately (#3495) @alexbarghi-nv
- [ENH] Add missing include of thrust/optional.h (#3493) @ahendriksen
- Remove wheel pytest verbosity (#3492) @sevagh
- Update clang-format to 16.0.1. (#3485) @bdice
- Use ARC V2 self-hosted runners for GPU jobs (#3483) @jjacobelli
- packed bool specialization to store edge endpoint|edge properties (#3482) @seunghwak
- Remove legacy renumber and shuffle calls from cython.cu (#3467) @ChuckHastings
- Remove legacy implementation of induce subgraph (#3464) @ChuckHastings
- Remove uses-setup-env-vars (#3463) @vyasr
- Optimize random walks (#3460) @jnke2016
- Update select_random_vertices to sample from a given distributed set or from (0, V] (#3455) @naimnv
v23.06.01
π¨ Breaking Changes
- [BUG] Fix Incorrect File Selection in cuGraph-PyG Loader (#3599) @alexbarghi-nv
- Remove legacy leiden (#3581) @ChuckHastings
- [IMP] Match Default PyG Hop ID Behavior in cuGraph-PyG (#3565) @alexbarghi-nv
- [IMP] Sample with Offsets in the Bulk Sampler (#3524) @alexbarghi-nv
- Dropping Python 3.8 (#3505) @divyegala
- Remove legacy renumber and shuffle calls from cython.cu (#3467) @ChuckHastings
- Remove legacy implementation of induce subgraph (#3464) @ChuckHastings
π Bug Fixes
- [HOTFIX] Fix random walks error identified in CUDA-12 testing (#3647) @ChuckHastings
- Fix MG Test Failing due to Removal of np.float (#3621) @alexbarghi-nv
- fix logic for shuffling results (#3619) @ChuckHastings
- [BUG] Fix Calls to cudf.DataFrame/Series.unique that relied on old behavior (#3616) @alexbarghi-nv
- correct dgl version in
cugraph-dgl
conda recipe (#3612) @tingyu66 - [BUG] Fix Issue in cuGraph-PyG Tests Blocking CI (#3607) @alexbarghi-nv
- [BUG] Critical: Fix cuGraph-PyG Edge Index Renumbering for Single-Edge Graphs (#3605) @alexbarghi-nv
- [BUG] Skip Empty Partitions in Bulk Sample Writing (#3600) @alexbarghi-nv
- [BUG] Fix Incorrect File Selection in cuGraph-PyG Loader (#3599) @alexbarghi-nv
- Fix SSSP bug (#3597) @jnke2016
- update cudf column constructor calls (#3592) @ChuckHastings
- Fix one more path to cugraphops in build workflow (#3554) @vyasr
- Fix path to cugraphops in build workflow (#3547) @vyasr
- Update dgl APIs for v1.1.0 (#3546) @tingyu66
- Pin to scikit-build<17.2 (#3538) @vyasr
- Correct results from sampling when grouping batches on specific GPUs (#3517) @ChuckHastings
- [FIX] Match the PyG API for Node Input to the Loader (#3514) @alexbarghi-nv
- Correct MG Leiden and SCC tests (#3509) @ChuckHastings
- per_v_transform_reduce_incoming|outgoing_e bug fix (when we're using (key, value) pairs to store edge src|dst property values) (#3508) @seunghwak
- Updates to allow python benchmarks to run on additional datasets by default (#3506) @rlratzel
- [BUG] Fix Intermittent Error when Converting cuDF DataFrame to Tensor by Converting to cuPy Array First (#3498) @alexbarghi-nv
- [FIX] Update cugraph-PyG Dependencies to include cuGraph (#3497) @alexbarghi-nv
- Fix graph_properties_t's members order (#3484) @naimnv
- Fix issue with latest rapids-make (#3481) @ChuckHastings
- Branch 23.06 Fix Forward Merge (#3462) @alexbarghi-nv
- Update raft dependency to 23.06 (#3410) @ChuckHastings
π Documentation
- updated cugraph Demo notebooks for 23.06 (#3558) @acostadon
- cugraph-ops license (#3553) @BradReesWork
- Notebook clean-up and run verification (#3551) @acostadon
- Updates contributing steps to add copyright and license text inclusion instruction (#3519) @rlratzel
- Fixed notebook links in algorithm and cugraph notebook pages (#3515) @acostadon
- adding cugraph-ops (#3488) @BradReesWork
- Sphinx updates (#3468) @BradReesWork
π New Features
- [REVIEW] Add MNMG with training (#3603) @VibhuJawa
- MG Leiden and MG MIS (#3582) @naimnv
- graph primitive transform_e (#3548) @seunghwak
- Support CUDA 12.0 for pip wheels (#3544) @divyegala
- Updates pytest benchmarks to use synthetic data and multi-GPUs (#3540) @rlratzel
- Enable edge masking (#3522) @seunghwak
- [REVIEW] Profile graph creation runtime and memory footprint (#3518) @VibhuJawa
- Bipartite R-mat graph generation. (#3512) @seunghwak
- Dropping Python 3.8 (#3505) @divyegala
- Creates Notebook that runs Multi-GPU versions of Jaccard, Sorenson and overlap. (#3504) @acostadon
- [cugraph-dgl] Add support for bipartite node features and optional edge features in GATConv (#3503) @tingyu66
- [cugraph-dgl] Add TransformerConv (#3501) @tingyu66
- [cugraph-pyg] Add TransformerConv and support for bipartite node features in GATConv (#3489) @tingyu66
- Branch 23.06 resolve merge conflict for forward merge (#3409) @alexbarghi-nv
- Refactor Leiden (#3327) @jnke2016
π οΈ Improvements
- Disables memory usage assertions (#3641) @rlratzel
- Refresh requirements (#3622) @jakirkham
- Pr3266 continue (optional arg for weight attribute for Nx graphs in
sssp
) (#3611) @eriknw - Enables MG python tests using a single-GPU LocalCUDACluster in CI (#3596) @rlratzel
- UVM notebook update and add tracker for notebooks to readme (#3595) @acostadon
- [REVIEW] Skip adding edge types, edge weights (#3583) @VibhuJawa
- Remove legacy leiden (#3581) @ChuckHastings
- run docs nightly too (#3568) @AyodeAwe
- include hop as part of the sort criteria for sampling results (#3567) @ChuckHastings
- Add MG python implementation of Leiden (#3566) @jnke2016
- [IMP] Match Default PyG Hop ID Behavior in cuGraph-PyG (#3565) @alexbarghi-nv
- Switch back to using primary shared-action-workflows branch (#3562) @vyasr
- removed deprecated calls and modified demo notebooks to run with 23.06 (#3561) @acostadon
- add unit test for checking is_symmetric is valid, update documentatio⦠(#3559) @ChuckHastings
- Update recipes to GTest version >=1.13.0 (#3549) @bdice
- Improve memory footprint and performance of graph creation (#3542) @VibhuJawa
- Update cupy dependency (#3539) @vyasr
- Perform expensive edge list check in create_graph_from_edgelist() (#3533) @seunghwak
- Enable sccache hits from local builds (#3526) @AyodeAwe
- Build wheels using new single image workflow (#3525) @vyasr
- [IMP] Sample with Offsets in the Bulk Sampler (#3524) @alexbarghi-nv
- Revert shared-action-workflows pin (#3523) @divyegala
- [FIX] fix cugraphops namespace (#3520) @stadlmax
- Add support in C API for handling unweighted graphs in algorithms that expect weights (#3513) @ChuckHastings
- Changes to support gtest version 1.11 (#3511) @ChuckHastings
- update docs (#3510) @BradReesWork
- Remove usage of rapids-get-rapids-version-from-git (#3502) @jjacobelli
- Remove Dummy Edge Weights, Support Specifying Edge Ids/Edge Types/Weights Separately (#3495) @alexbarghi-nv
- [ENH] Add missing include of thrust/optional.h (#3493) @ahendriksen
- Remove wheel pytest verbosity (#3492) @sevagh
- Update clang-format to 16.0.1. (#3485) @bdice
- Use ARC V2 self-hosted runners for GPU jobs (#3483) @jjacobelli
- packed bool specialization to store edge endpoint|edge properties (#3482) @seunghwak
- Remove legacy renumber and shuffle calls from cython.cu (#3467) @ChuckHastings
- Remove legacy implementation of induce subgraph (#3464) @ChuckHastings
- Remove uses-setup-env-vars (#3463) @vyasr
- Optimize random walks (#3460) @jnke2016
- Update select_random_vertices to sample from a given distributed set or from (0, V] (#3455) @naimnv
v23.06.00
π¨ Breaking Changes
- [BUG] Fix Incorrect File Selection in cuGraph-PyG Loader (#3599) @alexbarghi-nv
- Remove legacy leiden (#3581) @ChuckHastings
- [IMP] Match Default PyG Hop ID Behavior in cuGraph-PyG (#3565) @alexbarghi-nv
- [IMP] Sample with Offsets in the Bulk Sampler (#3524) @alexbarghi-nv
- Dropping Python 3.8 (#3505) @divyegala
- Remove legacy renumber and shuffle calls from cython.cu (#3467) @ChuckHastings
- Remove legacy implementation of induce subgraph (#3464) @ChuckHastings
π Bug Fixes
- Fix MG Test Failing due to Removal of np.float (#3621) @alexbarghi-nv
- fix logic for shuffling results (#3619) @ChuckHastings
- [BUG] Fix Calls to cudf.DataFrame/Series.unique that relied on old behavior (#3616) @alexbarghi-nv
- correct dgl version in
cugraph-dgl
conda recipe (#3612) @tingyu66 - [BUG] Fix Issue in cuGraph-PyG Tests Blocking CI (#3607) @alexbarghi-nv
- [BUG] Critical: Fix cuGraph-PyG Edge Index Renumbering for Single-Edge Graphs (#3605) @alexbarghi-nv
- [BUG] Skip Empty Partitions in Bulk Sample Writing (#3600) @alexbarghi-nv
- [BUG] Fix Incorrect File Selection in cuGraph-PyG Loader (#3599) @alexbarghi-nv
- Fix SSSP bug (#3597) @jnke2016
- update cudf column constructor calls (#3592) @ChuckHastings
- Fix one more path to cugraphops in build workflow (#3554) @vyasr
- Fix path to cugraphops in build workflow (#3547) @vyasr
- Update dgl APIs for v1.1.0 (#3546) @tingyu66
- Pin to scikit-build<17.2 (#3538) @vyasr
- Correct results from sampling when grouping batches on specific GPUs (#3517) @ChuckHastings
- [FIX] Match the PyG API for Node Input to the Loader (#3514) @alexbarghi-nv
- Correct MG Leiden and SCC tests (#3509) @ChuckHastings
- per_v_transform_reduce_incoming|outgoing_e bug fix (when we're using (key, value) pairs to store edge src|dst property values) (#3508) @seunghwak
- Updates to allow python benchmarks to run on additional datasets by default (#3506) @rlratzel
- [BUG] Fix Intermittent Error when Converting cuDF DataFrame to Tensor by Converting to cuPy Array First (#3498) @alexbarghi-nv
- [FIX] Update cugraph-PyG Dependencies to include cuGraph (#3497) @alexbarghi-nv
- Fix graph_properties_t's members order (#3484) @naimnv
- Fix issue with latest rapids-make (#3481) @ChuckHastings
- Branch 23.06 Fix Forward Merge (#3462) @alexbarghi-nv
- Update raft dependency to 23.06 (#3410) @ChuckHastings
π Documentation
- updated cugraph Demo notebooks for 23.06 (#3558) @acostadon
- cugraph-ops license (#3553) @BradReesWork
- Notebook clean-up and run verification (#3551) @acostadon
- Updates contributing steps to add copyright and license text inclusion instruction (#3519) @rlratzel
- Fixed notebook links in algorithm and cugraph notebook pages (#3515) @acostadon
- adding cugraph-ops (#3488) @BradReesWork
- Sphinx updates (#3468) @BradReesWork
π New Features
- [REVIEW] Add MNMG with training (#3603) @VibhuJawa
- MG Leiden and MG MIS (#3582) @naimnv
- graph primitive transform_e (#3548) @seunghwak
- Support CUDA 12.0 for pip wheels (#3544) @divyegala
- Updates pytest benchmarks to use synthetic data and multi-GPUs (#3540) @rlratzel
- Enable edge masking (#3522) @seunghwak
- [REVIEW] Profile graph creation runtime and memory footprint (#3518) @VibhuJawa
- Bipartite R-mat graph generation. (#3512) @seunghwak
- Dropping Python 3.8 (#3505) @divyegala
- Creates Notebook that runs Multi-GPU versions of Jaccard, Sorenson and overlap. (#3504) @acostadon
- [cugraph-dgl] Add support for bipartite node features and optional edge features in GATConv (#3503) @tingyu66
- [cugraph-dgl] Add TransformerConv (#3501) @tingyu66
- [cugraph-pyg] Add TransformerConv and support for bipartite node features in GATConv (#3489) @tingyu66
- Branch 23.06 resolve merge conflict for forward merge (#3409) @alexbarghi-nv
- Refactor Leiden (#3327) @jnke2016
π οΈ Improvements
- Disables memory usage assertions (#3641) @rlratzel
- Refresh requirements (#3622) @jakirkham
- Pr3266 continue (optional arg for weight attribute for Nx graphs in
sssp
) (#3611) @eriknw - Enables MG python tests using a single-GPU LocalCUDACluster in CI (#3596) @rlratzel
- UVM notebook update and add tracker for notebooks to readme (#3595) @acostadon
- [REVIEW] Skip adding edge types, edge weights (#3583) @VibhuJawa
- Remove legacy leiden (#3581) @ChuckHastings
- run docs nightly too (#3568) @AyodeAwe
- include hop as part of the sort criteria for sampling results (#3567) @ChuckHastings
- Add MG python implementation of Leiden (#3566) @jnke2016
- [IMP] Match Default PyG Hop ID Behavior in cuGraph-PyG (#3565) @alexbarghi-nv
- Switch back to using primary shared-action-workflows branch (#3562) @vyasr
- removed deprecated calls and modified demo notebooks to run with 23.06 (#3561) @acostadon
- add unit test for checking is_symmetric is valid, update documentatio⦠(#3559) @ChuckHastings
- Update recipes to GTest version >=1.13.0 (#3549) @bdice
- Improve memory footprint and performance of graph creation (#3542) @VibhuJawa
- Update cupy dependency (#3539) @vyasr
- Perform expensive edge list check in create_graph_from_edgelist() (#3533) @seunghwak
- Enable sccache hits from local builds (#3526) @AyodeAwe
- Build wheels using new single image workflow (#3525) @vyasr
- [IMP] Sample with Offsets in the Bulk Sampler (#3524) @alexbarghi-nv
- Revert shared-action-workflows pin (#3523) @divyegala
- [FIX] fix cugraphops namespace (#3520) @stadlmax
- Add support in C API for handling unweighted graphs in algorithms that expect weights (#3513) @ChuckHastings
- Changes to support gtest version 1.11 (#3511) @ChuckHastings
- update docs (#3510) @BradReesWork
- Remove usage of rapids-get-rapids-version-from-git (#3502) @jjacobelli
- Remove Dummy Edge Weights, Support Specifying Edge Ids/Edge Types/Weights Separately (#3495) @alexbarghi-nv
- [ENH] Add missing include of thrust/optional.h (#3493) @ahendriksen
- Remove wheel pytest verbosity (#3492) @sevagh
- Update clang-format to 16.0.1. (#3485) @bdice
- Use ARC V2 self-hosted runners for GPU jobs (#3483) @jjacobelli
- packed bool specialization to store edge endpoint|edge properties (#3482) @seunghwak
- Remove legacy renumber and shuffle calls from cython.cu (#3467) @ChuckHastings
- Remove legacy implementation of induce subgraph (#3464) @ChuckHastings
- Remove uses-setup-env-vars (#3463) @vyasr
- Optimize random walks (#3460) @jnke2016
- Update select_random_vertices to sample from a given distributed set or from (0, V] (#3455) @naimnv
[NIGHTLY] v23.08.00
π Links
π¨ Breaking Changes
- Change the renumber_sampled_edgelist function behavior. (#3762) @seunghwak
- PLC and Python Support for Sample-Side MFG Creation (#3734) @alexbarghi-nv
- Stop using setup.py in build.sh (#3704) @vyasr
- Refactor edge betweenness centrality (#3672) @jnke2016
- [FIX] Fix the hang in cuGraph Python Uniform Neighbor Sample, Add Logging to Bulk Sampler (#3669) @alexbarghi-nv
π Bug Fixes
- Makes copy of input ddf to work around dropped column names (#3776) @jnke2016
- Change the renumber_sampled_edgelist function behavior. (#3762) @seunghwak
- Fix bug discovered in Jaccard testing (#3758) @ChuckHastings
- fix inconsistent graph properties between the SG and the MG API (#3757) @jnke2016
- Fixes options for
--pydevelop
to remove unneeded CWD path ("."), restores use ofsetup.py
temporarily for develop builds (#3747) @rlratzel - Fix sampling call parameters if compiled with -DNO_CUGRAPH_OPS (#3729) @ChuckHastings
- Fix primitive bug discovered in MG edge betweenness centrality testing (#3723) @ChuckHastings
- Reorder dependencies.yaml channels (#3721) @raydouglass
- [BUG] Fix namesapce to default_hash and hash_functions (#3711) @naimnv
- [BUG] Fix Bulk Sampling Test Issue (#3701) @alexbarghi-nv
- Make
pylibcugraphops
optional imports incugraph-dgl
and-pyg
(#3693) @tingyu66 - [FIX] Rename
cugraph-ops
symbols (refactoring) and update GHA workflows to call pytest viapython -m pytest
(#3688) @naimnv - [FIX] Fix the hang in cuGraph Python Uniform Neighbor Sample, Add Logging to Bulk Sampler (#3669) @alexbarghi-nv
- force atlas notebook changes to run in cugraph 23.08 container. (#3656) @acostadon
π Documentation
- this fixes github links in cugraph, cugraph-dgl and cugraph-pyg (#3650) @acostadon
- Fix minor typo in README.md (#3636) @akasper
- Created landing spot for centrality and similarity algorithms (#3620) @acostadon
π New Features
- Compute shortest distances between given sets of origins and destinations for large diameter graphs (#3741) @seunghwak
- Update primitive to compute weighted Jaccard, Sorensen and Overlap similarity (#3728) @naimnv
- Add CUDA 12.0 conda environment. (#3725) @bdice
- Renumber utility function for sampling output (#3707) @seunghwak
- Integrate C++ Sampling Source Behavior Updates (#3699) @alexbarghi-nv
- Adds
fail_on_nonconvergence
option topagerank
to provide pagerank results even on non-convergence (#3639) @rlratzel - Add Benchmark for Bulk Sampling (#3628) @alexbarghi-nv
- cugraph: Build CUDA 12 packages (#3456) @vyasr
π οΈ Improvements
- Pin
dask
anddistributed
for23.08
release (#3761) @galipremsagar - Fix
build.yaml
workflow (#3756) @ajschmidt8 - Support MFG creation on sampling gpus for cugraph dgl (#3742) @VibhuJawa
- PLC and Python Support for Sample-Side MFG Creation (#3734) @alexbarghi-nv
- Switch to new wheel building pipeline (#3731) @vyasr
- Remove RAFT specialization. (#3727) @bdice
- C API for renumbering the samples (#3724) @ChuckHastings
- Only run cugraph conda CI for CUDA 11. (#3713) @bdice
- Promote
Datasets
to stable and clean-up unit tests (#3712) @nv-rliu - [BUG] Unsupported graph for similiarity algos (#3710) @jnke2016
- Stop using setup.py in build.sh (#3704) @vyasr
- [WIP] Make edge ids optional (#3702) @VibhuJawa
- Use rapids-cmake testing to run tests in parallel (#3697) @robertmaynard
- Sampling modifications to support PyG and DGL options (#3696) @ChuckHastings
- Include cuCollection public header for hash functions (#3694) @seunghwak
- Refactor edge betweenness centrality (#3672) @jnke2016
- Refactor RMAT (#3662) @jnke2016
- [REVIEW] Optimize bulk sampling (#3661) @VibhuJawa
- Update to CMake 3.26.4 (#3648) @vyasr
- Optimize cugraph-dgl MFG creation (#3646) @VibhuJawa
- use rapids-upload-docs script (#3640) @AyodeAwe
- Fix dependency versions for
23.08
(#3638) @ajschmidt8 - Unpin
dask
anddistributed
for development (#3634) @galipremsagar - Remove documentation build scripts for Jenkins (#3627) @ajschmidt8
- Unpin scikit-build upper bound (#3609) @vyasr
- Implement C++ Edge Betweenness Centrality (#3602) @ChuckHastings