Add cuTENSOR support & bug fixes discovered while working on conda testing by rwgk · Pull Request #1194 · NVIDIA/cuda-python · GitHub

Closes #1144, #1116

Bump cuda-pathfinder version to 1.3.2

Main changes:

find_nvidia_headers.py generalization for non-CTK headers, introducing a new family of SUPPORTED_*NON_CTK* variables in supported_nvidia_headers.py.
test_load_nvidia_dynamic_lib.py now loops consistently over all SUPPORTED_LINUX_SONAMES or SUPPORTED_WINDOWS_DLLS, depending on the platform.

Piggy-backed changes:

Fix [BUG]: cublasmp dependencies are not reflected in supported_nvidia_libs.py #1116
Add SITE_PACKAGES_LIBDIRS_WINDOWS_OTHER cudss paths in supported_nvidia_libs.py.
Handle cccl IS_WINDOWS conda anomaly (see this comment) in find_nvidia_headers.py (this was overlooked before due to lack of automatic testing).
commit 0d71a85 — Add nvidia-cufftmp-cu13 data in supported_nvidia_libs.py

… testing is INCOMPLETE.

…inder_testing.*

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

/ok to test

…test_load_nvidia_dynamic_lib.py

…or conda anomaly, to restore proper functioning for standard CTK installations

/ok to test

Manually tested with: pip install nvidia-cufftmp-cu13==12.1.3.1 That wheel was yanked, therefore not adding to pyproject.toml

Auto-sync is disabled for ready for review pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

/ok to test

Greptile Overview

{"summary": "

Sequence Diagram

sequenceDiagram
    participant User
    participant find_nvidia_header_directory
    participant find_nvidia_headers.py
    participant supported_nvidia_headers.py
    participant load_nvidia_dynamic_lib
    participant supported_nvidia_libs.py
    participant Test Suite

    User->>find_nvidia_header_directory: "Request header for libname (e.g., 'cutensor')"
    find_nvidia_header_directory->>supported_nvidia_headers.py: "Check if libname in SUPPORTED_HEADERS_CTK"
    alt CTK Library
        supported_nvidia_headers.py-->>find_nvidia_header_directory: "Found in CTK"
        find_nvidia_header_directory->>find_nvidia_headers.py: "Call _find_ctk_header_directory()"
    else Non-CTK Library (e.g., cutensor)
        supported_nvidia_headers.py-->>find_nvidia_header_directory: "Check SUPPORTED_HEADERS_NON_CTK"
        find_nvidia_header_directory->>supported_nvidia_headers.py: "Get canonical header basename"
        supported_nvidia_headers.py-->>find_nvidia_header_directory: "Return 'cutensor.h'"
        find_nvidia_header_directory->>find_nvidia_headers.py: "Search site-packages"
        find_nvidia_headers.py->>find_nvidia_headers.py: "Check SUPPORTED_SITE_PACKAGE_HEADER_DIRS_NON_CTK"
        find_nvidia_headers.py->>find_nvidia_headers.py: "_find_based_on_conda_layout() for non-CTK"
        alt Windows Conda Anomaly (cccl)
            find_nvidia_headers.py->>find_nvidia_headers.py: "Handle targets/x64 path anomaly"
        end
        find_nvidia_headers.py->>find_nvidia_headers.py: "Check SUPPORTED_INSTALL_DIRS_NON_CTK"
    end
    find_nvidia_header_directory-->>User: "Return header directory path or None"

    User->>load_nvidia_dynamic_lib: "Load dynamic library (e.g., 'cutensor')"
    load_nvidia_dynamic_lib->>supported_nvidia_libs.py: "Query SUPPORTED_LINUX_SONAMES or SUPPORTED_WINDOWS_DLLS"
    alt CTK Library
        supported_nvidia_libs.py-->>load_nvidia_dynamic_lib: "Return from SUPPORTED_*_SONAMES_CTK"
    else Non-CTK Library (cutensor)
        supported_nvidia_libs.py-->>load_nvidia_dynamic_lib: "Return from SUPPORTED_*_SONAMES_OTHER"
        Note over supported_nvidia_libs.py: "Added: cutensor -> libcutensor.so.2 (Linux)<br/>cutensor -> cutensor.dll (Windows)"
    end
    load_nvidia_dynamic_lib->>supported_nvidia_libs.py: "Check SITE_PACKAGES_LIBDIRS_*"
    supported_nvidia_libs.py-->>load_nvidia_dynamic_lib: "Return site-packages paths"
    load_nvidia_dynamic_lib-->>User: "Return LoadedDL object or DynamicLibNotFoundError"

    User->>Test Suite: "Run test_find_nvidia_headers"
    Test Suite->>find_nvidia_header_directory: "Test all SUPPORTED_HEADERS_NON_CTK.keys()"
    Test Suite->>Test Suite: "Loop over cutensor, nvshmem, etc."
    Test Suite-->>User: "Report test results"

    User->>Test Suite: "Run test_load_nvidia_dynamic_lib"
    Test Suite->>load_nvidia_dynamic_lib: "Test all SUPPORTED_LINUX_SONAMES or SUPPORTED_WINDOWS_DLLS"
    Test Suite->>Test Suite: "Run in spawned child process for isolation"
    Test Suite-->>User: "Report test results with abs_path or 'Not found'"

14 files reviewed, 12 comments

Edit Code Review Agent Settings | Greptile

@ZzEeKkAa I believe this is ready for review. It'd be great if you could take a look.

The only thing left to do is go through the manual testing systematically.

CI ran successfully at commit b079c84:

https://github.com/NVIDIA/cuda-python/actions/runs/18886735456/job/53903916554?pr=1194

Greptile Overview

Greptile Summary

This incremental review covers only the changes made since the last review, not the entire PR. The developer has addressed previous feedback by applying the _abs_norm() wrapper to all remaining return paths in find_nvidia_header_directory() (lines 150, 153, 159) and fixed the PowerShell syntax error in the conda setup script. The _abs_norm() helper (lines 15-18) normalizes path separators and converts relative paths to absolute paths, ensuring consistent path format across all return points—critical for the new cuTENSOR/non-CTK library support that can be installed in diverse locations (site-packages, conda, standard directories). The PowerShell script now includes 'cutensor' in the package list for testing and corrects the missing comma after "libnvshmem-dev". These changes ensure that header discovery returns predictable, normalized paths regardless of installation method or platform, addressing Windows conda path anomalies mentioned in the PR description.

Important Files Changed

Filename	Score	Overview
cuda_pathfinder/cuda/pathfinder/_headers/find_nvidia_headers.py	5/5	Applied _abs_norm() wrapper to all non-CTK header return paths for consistent path normalization
toolshed/conda_create_for_pathfinder_testing.ps1	5/5	Added 'cutensor' package and fixed trailing comma syntax error in package list

Confidence score: 5/5

This PR is safe to merge with minimal risk as the changes are targeted fixes addressing specific previous review feedback
Score reflects that all previously identified issues have been resolved: path normalization is now consistent across all return paths, and the PowerShell syntax error has been corrected
No files require special attention; both changes are straightforward defensive improvements that enhance cross-platform reliability

2 files reviewed, no comments

Edit Code Review Agent Settings | Greptile

/ok to test

…headers.py (mirrors supported_nvidia_libs.py)

I'm through one pass of running all 18 combinations of manual tests. I'll log the results here tomorrow.

Manual QA, based on local tests following this scheme:

Results based on the logs in the attached pr1194_local_testing_pass01_2025-10-29+094920.zip

local-ctk headers (note that these are the only matches for the grep)

pr1194_local_testing/pass01 $ grep 'hdr_dir=None' *.txt | sort -t'[' -k2
2lc_1lx64_cu12.txt:INFO test_find_non_ctk_headers[cutensor]: hdr_dir=None
2lc_1lx64_cu13.txt:INFO test_find_non_ctk_headers[cutensor]: hdr_dir=None
2lc_2la64_cu12.txt:INFO test_find_non_ctk_headers[cutensor]: hdr_dir=None
2lc_2la64_cu13.txt:INFO test_find_non_ctk_headers[cutensor]: hdr_dir=None
2lc_3wx64_cu12.txt:INFO test_find_non_ctk_headers[cutensor]: hdr_dir=None
2lc_3wx64_cu13.txt:INFO test_find_non_ctk_headers[cutensor]: hdr_dir=None
2lc_2la64_cu12.txt:INFO test_find_non_ctk_headers[nvshmem]: hdr_dir=None
2lc_2la64_cu13.txt:INFO test_find_non_ctk_headers[nvshmem]: hdr_dir=None

Explanation: cutensor and nvshmem are not installed when testing with local-ctk.

site-packages libs

pr1194_local_testing/pass01 $ grep 'Not found:' 1sp_*.txt | sort -t"'" -k2
1sp_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
1sp_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
1sp_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
1sp_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
1sp_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
1sp_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'
1sp_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'

Explanations:

cufftMp and mathdx are not available for cu13. (freshly verified by visiting pypi.org)
nvpl_fftw is available only for linux-aarch64.

conda-ctk libs

pr1194_local_testing/pass01 $ grep 'Not found:' 3cc_*.txt | sort -t"'" -k2
3cc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
3cc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
3cc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'
3cc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'

Explanations:

cufftMp is not available for cu13.
nvpl_fftw is available only for linux-aarch64.

local-ctk libs

pr1194_local_testing/pass01 $ grep 'Not found:' 2lc_*.txt | sort -t"'" -k2
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cublasmp]: Not found: libname='cublasmp'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cublasmp]: Not found: libname='cublasmp'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cublasmp]: Not found: libname='cublasmp'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cublasmp]: Not found: libname='cublasmp'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cudss]: Not found: libname='cudss'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cudss]: Not found: libname='cudss'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cudss]: Not found: libname='cudss'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cudss]: Not found: libname='cudss'
2lc_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cudss]: Not found: libname='cudss'
2lc_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cudss]: Not found: libname='cudss'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cutensor]: Not found: libname='cutensor'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cutensor]: Not found: libname='cutensor'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cutensor]: Not found: libname='cutensor'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cutensor]: Not found: libname='cutensor'
2lc_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cutensor]: Not found: libname='cutensor'
2lc_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cutensor]: Not found: libname='cutensor'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cutensorMg]: Not found: libname='cutensorMg'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cutensorMg]: Not found: libname='cutensorMg'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cutensorMg]: Not found: libname='cutensorMg'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cutensorMg]: Not found: libname='cutensorMg'
2lc_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cutensorMg]: Not found: libname='cutensorMg'
2lc_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cutensorMg]: Not found: libname='cutensorMg'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
2lc_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
2lc_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nccl]: Not found: libname='nccl'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nccl]: Not found: libname='nccl'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nccl]: Not found: libname='nccl'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nccl]: Not found: libname='nccl'
2lc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'
2lc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nvpl_fftw]: Not found: libname='nvpl_fftw'
2lc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[nvshmem_host]: Not found: libname='nvshmem_host'
2lc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[nvshmem_host]: Not found: libname='nvshmem_host'

Explanation: These are all non-CTK libs; only the local-ctk is installed.

A special look at cufftMp:

pr1194_local_testing/pass01 $ grep 'INFO test_load_nvidia_dynamic_lib\[cufftMp\]' 1sp_*.txt 3cc_*.txt
1sp_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: abs_path=/home/rgrossekunst/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cufftmp/cu12/lib/libcufftMp.so.11
1sp_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
1sp_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: abs_path=/wrk/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cufftmp/cu12/lib/libcufftMp.so.11
1sp_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'

3cc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: abs_path=/home/rgrossekunst/miniforge3/envs/pathfinder_cu12.9.1/lib/libcufftMp.so
3cc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'
3cc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: abs_path=/wrk/miniforge3/envs/pathfinder_testing_cu12.9.1/lib/libcufftMp.so
3cc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cufftMp]: Not found: libname='cufftMp'

Confirms the explanations above.

A special look at mathdx:

pr1194_local_testing/pass01 $ grep 'INFO test_load_nvidia_dynamic_lib\[mathdx\]' 1sp_*.txt 3cc_*.txt
1sp_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=/home/rgrossekunst/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cu12/lib/libmathdx.so.0
1sp_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
1sp_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=/wrk/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cu12/lib/libmathdx.so.0
1sp_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'
1sp_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=C:\Users\rgrossekunst\forked\cuda-python\Sp12Venv\Lib\site-packages\nvidia/cu12/bin\mathdx64_0.dll
1sp_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: Not found: libname='mathdx'

3cc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=/home/rgrossekunst/miniforge3/envs/pathfinder_cu12.9.1/lib/libmathdx.so
3cc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=/home/rgrossekunst/miniforge3/envs/pathfinder_testing_cu13.0.2/lib/libmathdx.so
3cc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=/wrk/miniforge3/envs/pathfinder_testing_cu12.9.1/lib/libmathdx.so
3cc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=/wrk/miniforge3/envs/pathfinder_testing_cu13.0.2/lib/libmathdx.so
3cc_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=C:\Users\rgrossekunst\AppData\Local\miniforge3\envs\pathfinder_testing_cu12.9.1\Library\bin\mathdx64_0.dll
3cc_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[mathdx]: abs_path=C:\Users\rgrossekunst\AppData\Local\miniforge3\envs\pathfinder_testing_cu13.0.2\Library\bin\mathdx64_0.dll

Unexplained: mathdx for cu13 is available under conda, but not site-packages

pr1194_local_testing_pass01_2025-10-29+094920.zip

@leofang Could you please review the new #1194 (comment)?

Unless you find an issue there, I believe this PR is ready for merging.

Regarding the "conda && windows && cccl" anomaly: Possibly the implementation could be streamlined, by reworking the anchor_point logic, but what we have here is a strict improvement (because the logic was completely missing before), and polishing the implementation will only be a very minor detail; this surely isn't structural on the pathfinder side.

/ok to test 03e10b1

FWIW (it doesn't really matter for this PR), I looked into this for a few minutes:

Unexplained: mathdx for cu13 is available under conda, but not site-packages

https://chatgpt.com/share/69025225-d368-8008-9c03-80a09953de70

Bottom line: That's just the way it is right now.

… offline

/ok to test

Regarding commit 43e81b4, I logged my thoughts here:

#1197 (comment)

Thanks @leofang!

@ZzEeKkAa I'll work on the pathfinder 1.3.2 release straightaway.

Doc Preview CI
Preview removed because the pull request was closed or merged.

rwgk added 10 commits October 26, 2025 22:49

Add support for cutensor. Still works for existing libnames, cutensor…

2d1d8fe

… testing is INCOMPLETE.

Generalize tests/test_find_nvidia_headers.py to also cover cutensor

986a83f

test_find_nvidia_headers.py conda testing and fix

0583393

test_load_nvidia_dynamic_lib.py fix conda testing

1d1f534

Add conda_create_for_pathfinder_testing.ps1

e0a4ca6

Bug fix: SITE_PACKAGES_LIBDIRS_WINDOWS_OTHER cutensor, cutensorMg paths

25dd364

Add cudss paths to SITE_PACKAGES_LIBDIRS_WINDOWS_OTHER

cdab969

Add SUPPORTED_HEADERS_NON_CTK_ALL to fix Windows site-packages tests

6d490f5

Bug fix (existing code): conda cccl header directory

28e7206

test_find_nvidia_headers.py: refer to toolshed/conda_create_for_pathf…

d36a62d

…inder_testing.*

This comment has been minimized.

Sign in to view

rwgk added 3 commits October 28, 2025 09:09

nvidia-libmathdx-... only exists for cu12: tolerate abs_path=None in …

0734ac7

…test_load_nvidia_dynamic_lib.py

find_nvidia_headers.py cccl IS_WINDOWS: fall-through after checking f…

3b156b7

…or conda anomaly, to restore proper functioning for standard CTK installations

Add cublasmp DIRECT_DEPENDENCIES (closes NVIDIA#1116 )

7aa6679

rwgk added 4 commits October 28, 2025 10:57

Add SUPPORTED_HEADERS_NON_CTK to cuda_pathfinder/docs/source/api.rst

9bfef5f

Add 1.3.2-notes.rst

3a44b6a

Add nvidia-cufftmp-cu13 data in supported_nvidia_libs.py

0d71a85

Manually tested with: pip install nvidia-cufftmp-cu13==12.1.3.1 That wheel was yanked, therefore not adding to pyproject.toml

Merge branch 'main' into cutensor_support

b079c84

rwgk marked this pull request as ready for review October 28, 2025 19:23

greptile-apps Bot reviewed Oct 28, 2025

View reviewed changes

rwgk added 2 commits October 28, 2025 12:56

Add missing comma in toolshed/conda_create_for_pathfinder_testing.ps1

90ed27a

Systematically add _abs_norm() in find_nvidia_header_directory()

9671b24

greptile-apps Bot reviewed Oct 28, 2025

View reviewed changes

leofang added enhancement Any code-related improvements P0 High priority - Must do! cuda.pathfinder Everything related to the cuda.pathfinder module labels Oct 28, 2025

leofang reviewed Oct 28, 2025

View reviewed changes

Comment thread cuda_pathfinder/docs/source/api.rst

leofang reviewed Oct 28, 2025

View reviewed changes

Comment thread cuda_pathfinder/cuda/pathfinder/__init__.py

leofang reviewed Oct 28, 2025

View reviewed changes

Comment thread cuda_pathfinder/cuda/pathfinder/_headers/find_nvidia_headers.py Outdated

rwgk force-pushed the cutensor_support branch from ab4c8f3 to 9671b24 Compare October 28, 2025 22:13

leofang reviewed Oct 28, 2025

View reviewed changes

Comment thread cuda_pathfinder/cuda/pathfinder/_headers/supported_nvidia_headers.py Outdated

rwgk added 2 commits October 28, 2025 17:29

Move "conda has this anomaly" comment to the end of the line

4d5a41a

Merge branch 'main' into cutensor_support

75d8874

rwgk mentioned this pull request Oct 29, 2025

[ENH]: Clean up SUPPORTED_... variables between load libs and find header dirs #1197

Closed

leofang reviewed Oct 29, 2025

View reviewed changes

Comment thread cuda_pathfinder/tests/test_find_nvidia_headers.py

leofang reviewed Oct 29, 2025

View reviewed changes

Comment thread cuda_pathfinder/tests/test_load_nvidia_dynamic_lib.py

Add empty SUPPORTED_HEADERS_NON_CTK_WINDOWS_ONLY in supported_nvidia_…

03e10b1

…headers.py (mirrors supported_nvidia_libs.py)

leofang added this to the cuda.pathfinder 1.3.2 milestone Oct 29, 2025

Remove SUPPORTED_HEADERS_NON_CTK from public API, as requested by Leo…

43e81b4

… offline

leofang approved these changes Oct 29, 2025

View reviewed changes

rwgk merged commit 9769bc7 into NVIDIA:main Oct 29, 2025
64 checks passed

rwgk deleted the cutensor_support branch October 29, 2025 21:13

LeSingh1 mentioned this pull request Aug 4, 2026

Fix cufftMp soname priority: tabulate linux_sonames oldest-first #2499

Open

Add cuTENSOR support & bug fixes discovered while working on conda testing by rwgk · Pull Request #1194 · NVIDIA/cuda-python · GitHub

Uh oh!

Add cuTENSOR support & bug fixes discovered while working on conda testing - #1194

Add cuTENSOR support & bug fixes discovered while working on conda testing#1194rwgk merged 23 commits intoNVIDIA:mainNVIDIA/cuda-python:mainfrom rwgk:cutensor_supportrwgk/cuda-python:cutensor_supportCopy head branch name to clipboard

Conversation

rwgk commented Oct 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

copy-pr-bot Bot commented Oct 28, 2025

Uh oh!

rwgk commented Oct 28, 2025

Uh oh!

This comment has been minimized.

rwgk commented Oct 28, 2025

Uh oh!

copy-pr-bot Bot commented Oct 28, 2025

Uh oh!

rwgk commented Oct 28, 2025

Uh oh!

greptile-apps Bot left a comment

Choose a reason for hiding this comment

Greptile Overview

{"summary": "

Sequence Diagram

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rwgk commented Oct 28, 2025

Uh oh!

rwgk commented Oct 28, 2025

Uh oh!

greptile-apps Bot left a comment

Choose a reason for hiding this comment

Greptile Overview

Greptile Summary

Important Files Changed

Confidence score: 5/5

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

Uh oh!

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

leofang commented Oct 29, 2025

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

Uh oh!

rwgk commented Oct 29, 2025

Uh oh!

github-actions Bot commented Oct 29, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

Add cuTENSOR support & bug fixes discovered while working on conda testing#1194
rwgk merged 23 commits into
NVIDIA:mainfrom
rwgk:cutensor_support

rwgk commented Oct 28, 2025 •
edited
Loading