Skip to content

Conversation

@rwgk
Copy link
Collaborator

@rwgk rwgk commented Oct 30, 2025

Closes #1193

Bump cuda-pathfinder version to 1.3.3a0

Local manual testing passes with (site-packages, conda-ctk) x (linux-64, linux-aarch64, win-64) x (cu12, cu13) using the scheme shown here:

rwgk-win11.localdomain:~/col $ grep '^INFO test_.*: hdr_dir=' *.txt | tr -s '\\' '\\'
1sp_1lx64_cu12.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/home/rgrossekunst/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cusparselt/include'
1sp_1lx64_cu13.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/home/rgrossekunst/forked/cuda-python/SpCu13Venv/lib/python3.12/site-packages/nvidia/cusparselt/include'
1sp_2la64_cu12.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/wrk/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cusparselt/include'
1sp_2la64_cu13.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/wrk/forked/cuda-python/SpCu13Venv/lib/python3.12/site-packages/nvidia/cusparselt/include'
1sp_3wx64_cu12.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\Users\rgrossekunst\forked\cuda-python\Sp12Venv\Lib\site-packages\nvidia\cusparselt\include'
1sp_3wx64_cu13.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\Users\rgrossekunst\forked\cuda-python\Sp13Venv\Lib\site-packages\nvidia\cusparselt\include'
3cc_1lx64_cu12.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/home/rgrossekunst/miniforge3/envs/pathfinder_testing_cu12.9.1/include'
3cc_1lx64_cu13.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/home/rgrossekunst/miniforge3/envs/pathfinder_testing_cu13.0.2/include'
3cc_2la64_cu12.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/wrk/miniforge3/envs/pathfinder_testing_cu12.9.1/include'
3cc_2la64_cu13.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/wrk/miniforge3/envs/pathfinder_testing_cu13.0.2/include'
3cc_3wx64_cu12.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\Users\rgrossekunst\AppData\Local\miniforge3\envs\pathfinder_testing_cu12.9.1\Library\include'
3cc_3wx64_cu13.txt:INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\Users\rgrossekunst\AppData\Local\miniforge3\envs\pathfinder_testing_cu13.0.2\Library\include'
rwgk-win11.localdomain:~/col $ grep '^INFO test_.*: abs_path=' *.txt | tr -s '\\' '\\'
1sp_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/home/rgrossekunst/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
1sp_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/home/rgrossekunst/forked/cuda-python/SpCu13Venv/lib/python3.12/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
1sp_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/wrk/forked/cuda-python/SpCu12Venv/lib/python3.12/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
1sp_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/wrk/forked/cuda-python/SpCu13Venv/lib/python3.12/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
1sp_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\Users\rgrossekunst\forked\cuda-python\Sp12Venv\Lib\site-packages\nvidia/cusparselt/bin\cusparseLt.dll
1sp_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\Users\rgrossekunst\forked\cuda-python\Sp13Venv\Lib\site-packages\nvidia/cusparselt/bin\cusparseLt.dll
3cc_1lx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/home/rgrossekunst/miniforge3/envs/pathfinder_testing_cu12.9.1/lib/libcusparseLt.so
3cc_1lx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/home/rgrossekunst/miniforge3/envs/pathfinder_testing_cu13.0.2/lib/libcusparseLt.so
3cc_2la64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/wrk/miniforge3/envs/pathfinder_testing_cu12.9.1/lib/libcusparseLt.so
3cc_2la64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/wrk/miniforge3/envs/pathfinder_testing_cu13.0.2/lib/libcusparseLt.so
3cc_3wx64_cu12.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\Users\rgrossekunst\AppData\Local\miniforge3\envs\pathfinder_testing_cu12.9.1\Library\bin\cusparseLt.dll
3cc_3wx64_cu13.txt:INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\Users\rgrossekunst\AppData\Local\miniforge3\envs\pathfinder_testing_cu13.0.2\Library\bin\cusparseLt.dll

local-ctk testing is omitted, because

  • cusparseLt is not part of the CTK
  • we only support site-packages and conda-ctk for cusparseLt

Test logs: pr1200_testing_2025-10-29+213040.zip

@rwgk rwgk self-assigned this Oct 30, 2025
@rwgk rwgk added enhancement Any code-related improvements cuda.pathfinder Everything related to the cuda.pathfinder module labels Oct 30, 2025
@copy-pr-bot
Copy link
Contributor

copy-pr-bot bot commented Oct 30, 2025

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.

@rwgk
Copy link
Collaborator Author

rwgk commented Oct 30, 2025

/ok to test

@github-actions
Copy link

@rwgk
Copy link
Collaborator Author

rwgk commented Oct 30, 2025

@rwgk rwgk force-pushed the cusparselt_support branch from a333403 to ba1e17f Compare October 30, 2025 04:44
@rwgk rwgk marked this pull request as ready for review October 30, 2025 04:45
@copy-pr-bot
Copy link
Contributor

copy-pr-bot bot commented Oct 30, 2025

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.

@rwgk rwgk requested a review from leofang October 30, 2025 04:45
@rwgk
Copy link
Collaborator Author

rwgk commented Oct 30, 2025

Not rerunning the CI because the last commit (ba1e17f) only changed the pathfinder version number.

Copy link
Contributor

@cpcloud cpcloud left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🚢

@rwgk
Copy link
Collaborator Author

rwgk commented Oct 30, 2025

/ok to test

@rwgk rwgk enabled auto-merge (squash) October 30, 2025 19:35
@rwgk
Copy link
Collaborator Author

rwgk commented Oct 30, 2025

Quick manual QA based on

rwgk-win11.localdomain:~/logs_48738060654 $ grep 'Z INFO test_load_nvidia_dynamic_lib\[cusparseLt\]: abs_path=' *Test*.txt | tr -s '\\' '\\' | cut -d: -f1 | uniq
10_Test linux-aarch64 _ py3.12, 13.0.2, wheels, GPU a100.txt
11_Test linux-aarch64 _ py3.13, 12.9.1, wheels, GPU a100.txt
12_Test linux-aarch64 _ py3.14t, 13.0.2, local, GPU a100.txt
13_Test linux-aarch64 _ py3.11, 13.0.2, local, GPU a100.txt
14_Test linux-aarch64 _ py3.11, 12.9.1, wheels, GPU a100.txt
15_Test linux-aarch64 _ py3.13, 13.0.2, local, GPU a100.txt
16_Test linux-aarch64 _ py3.12, 12.9.1, local, GPU a100.txt
17_Test linux-aarch64 _ py3.14, 13.0.2, local, GPU a100.txt
18_Test linux-aarch64 _ py3.9, 12.9.1, wheels, GPU a100.txt
19_Test linux-aarch64 _ py3.10, 12.9.1, local, GPU a100.txt
1_Test win-64 _ py3.13, 13.0.2, local, GPU l4.txt
20_Test linux-aarch64 _ py3.10, 13.0.2, wheels, GPU a100.txt
21_Test linux-aarch64 _ py3.9, 13.0.2, local, GPU a100.txt
23_Test linux-64 _ py3.10, 12.9.1, local, GPU l4.txt
24_Test linux-64 _ py3.14t, 13.0.2, local, GPU l4.txt
25_Test linux-64 _ py3.12, 12.9.1, local, GPU l4.txt
26_Test linux-64 _ py3.9, 13.0.2, local, GPU l4.txt
27_Test linux-64 _ py3.13, 13.0.2, local, GPU l4.txt
28_Test linux-64 _ py3.13, 12.9.1, wheels, GPU l4.txt
29_Test linux-64 _ py3.14, 13.0.2, local, GPU l4.txt
2_Test win-64 _ py3.14t, 13.0.2, local, GPU l4.txt
30_Test linux-64 _ py3.13, 13.0.2, local, GPU H100.txt
31_Test linux-64 _ py3.11, 12.9.1, wheels, GPU l4.txt
32_Test linux-64 _ py3.10, 13.0.2, wheels, GPU l4.txt
33_Test linux-64 _ py3.11, 13.0.2, local, GPU l4.txt
34_Test linux-64 _ py3.12, 13.0.2, wheels, GPU l4.txt
35_Test linux-64 _ py3.9, 12.9.1, wheels, GPU l4.txt
3_Test win-64 _ py3.14, 13.0.2, wheels, GPU t4.txt
4_Test win-64 _ py3.14t, 13.0.2, wheels, GPU t4.txt
5_Test win-64 _ py3.12, 12.9.1, local, GPU t4.txt
6_Test win-64 _ py3.13, 13.0.2, wheels, GPU t4.txt
7_Test win-64 _ py3.14, 13.0.2, local, GPU l4.txt
8_Test win-64 _ py3.12, 12.9.1, wheels, GPU l4.txt
rwgk-win11.localdomain:~/logs_48738060654 $ grep 'Z INFO test_load_nvidia_dynamic_lib\[cusparseLt\]: abs_path=' *Test*.txt | tr -s '\\' '\\' | cut -d: -f2- | cut -dZ -f
2- | sort | uniq -c
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.10.19/arm64/lib/python3.10/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.10.19/x64/lib/python3.10/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.11.14/arm64/lib/python3.11/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.11.14/x64/lib/python3.11/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.12.12/arm64/lib/python3.12/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.12.12/x64/lib/python3.12/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.13.9/arm64/lib/python3.13/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      3  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.13.9/x64/lib/python3.13/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      1  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.14.0/arm64-freethreaded/lib/python3.14t/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      1  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.14.0/arm64/lib/python3.14/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      1  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.14.0/x64-freethreaded/lib/python3.14t/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      1  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.14.0/x64/lib/python3.14/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.9.24/arm64/lib/python3.9/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=/opt/hostedtoolcache/Python/3.9.24/x64/lib/python3.9/site-packages/nvidia/cusparselt/lib/libcusparseLt.so.0
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\actions-runner\_work\_tool\Python\3.12.10\x64\Lib\site-packages\nvidia/cusparselt/bin\cusparseLt.dll
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\actions-runner\_work\_tool\Python\3.13.9\x64\Lib\site-packages\nvidia/cusparselt/bin\cusparseLt.dll
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\actions-runner\_work\_tool\Python\3.14.0\x64-freethreaded\Lib\site-packages\nvidia/cusparselt/bin\cusparseLt.dll
      2  INFO test_load_nvidia_dynamic_lib[cusparseLt]: abs_path=C:\actions-runner\_work\_tool\Python\3.14.0\x64\Lib\site-packages\nvidia/cusparselt/bin\cusparseLt.dll

rwgk-win11.localdomain:~/logs_48738060654 $ grep 'Z INFO test_load_nvidia_dynamic_lib\[cusparseLt\]: abs_path=' *Test*.txt | tr -s '\\' '\\' | cut -d: -f1 | uniq
10_Test linux-aarch64 _ py3.12, 13.0.2, wheels, GPU a100.txt
11_Test linux-aarch64 _ py3.13, 12.9.1, wheels, GPU a100.txt
12_Test linux-aarch64 _ py3.14t, 13.0.2, local, GPU a100.txt
13_Test linux-aarch64 _ py3.11, 13.0.2, local, GPU a100.txt
14_Test linux-aarch64 _ py3.11, 12.9.1, wheels, GPU a100.txt
15_Test linux-aarch64 _ py3.13, 13.0.2, local, GPU a100.txt
16_Test linux-aarch64 _ py3.12, 12.9.1, local, GPU a100.txt
17_Test linux-aarch64 _ py3.14, 13.0.2, local, GPU a100.txt
18_Test linux-aarch64 _ py3.9, 12.9.1, wheels, GPU a100.txt
19_Test linux-aarch64 _ py3.10, 12.9.1, local, GPU a100.txt
1_Test win-64 _ py3.13, 13.0.2, local, GPU l4.txt
20_Test linux-aarch64 _ py3.10, 13.0.2, wheels, GPU a100.txt
21_Test linux-aarch64 _ py3.9, 13.0.2, local, GPU a100.txt
23_Test linux-64 _ py3.10, 12.9.1, local, GPU l4.txt
24_Test linux-64 _ py3.14t, 13.0.2, local, GPU l4.txt
25_Test linux-64 _ py3.12, 12.9.1, local, GPU l4.txt
26_Test linux-64 _ py3.9, 13.0.2, local, GPU l4.txt
27_Test linux-64 _ py3.13, 13.0.2, local, GPU l4.txt
28_Test linux-64 _ py3.13, 12.9.1, wheels, GPU l4.txt
29_Test linux-64 _ py3.14, 13.0.2, local, GPU l4.txt
2_Test win-64 _ py3.14t, 13.0.2, local, GPU l4.txt
30_Test linux-64 _ py3.13, 13.0.2, local, GPU H100.txt
31_Test linux-64 _ py3.11, 12.9.1, wheels, GPU l4.txt
32_Test linux-64 _ py3.10, 13.0.2, wheels, GPU l4.txt
33_Test linux-64 _ py3.11, 13.0.2, local, GPU l4.txt
34_Test linux-64 _ py3.12, 13.0.2, wheels, GPU l4.txt
35_Test linux-64 _ py3.9, 12.9.1, wheels, GPU l4.txt
3_Test win-64 _ py3.14, 13.0.2, wheels, GPU t4.txt
4_Test win-64 _ py3.14t, 13.0.2, wheels, GPU t4.txt
5_Test win-64 _ py3.12, 12.9.1, local, GPU t4.txt
6_Test win-64 _ py3.13, 13.0.2, wheels, GPU t4.txt
7_Test win-64 _ py3.14, 13.0.2, local, GPU l4.txt
8_Test win-64 _ py3.12, 12.9.1, wheels, GPU l4.txt
rwgk-win11.localdomain:~/logs_48738060654 $ grep 'Z INFO test_.*\[cusparseLt\]: hdr_dir=' *Test*.txt | grep -v ': hdr_dir=None' | tr -s '\\' '\\' | cut -d: -f2- | cut -dZ -f2- | sort | uniq -c
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.10.19/arm64/lib/python3.10/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.10.19/x64/lib/python3.10/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.11.14/arm64/lib/python3.11/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.11.14/x64/lib/python3.11/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.12.12/arm64/lib/python3.12/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.12.12/x64/lib/python3.12/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.13.9/arm64/lib/python3.13/site-packages/nvidia/cusparselt/include'
      3  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.13.9/x64/lib/python3.13/site-packages/nvidia/cusparselt/include'
      1  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.14.0/arm64-freethreaded/lib/python3.14t/site-packages/nvidia/cusparselt/include'
      1  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.14.0/arm64/lib/python3.14/site-packages/nvidia/cusparselt/include'
      1  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.14.0/x64-freethreaded/lib/python3.14t/site-packages/nvidia/cusparselt/include'
      1  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.14.0/x64/lib/python3.14/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.9.24/arm64/lib/python3.9/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='/opt/hostedtoolcache/Python/3.9.24/x64/lib/python3.9/site-packages/nvidia/cusparselt/include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\actions-runner\_work\_tool\Python\3.12.10\x64\Lib\site-packages\nvidia\cusparselt\include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\actions-runner\_work\_tool\Python\3.13.9\x64\Lib\site-packages\nvidia\cusparselt\include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\actions-runner\_work\_tool\Python\3.14.0\x64-freethreaded\Lib\site-packages\nvidia\cusparselt\include'
      2  INFO test_find_non_ctk_headers[cusparseLt]: hdr_dir='C:\actions-runner\_work\_tool\Python\3.14.0\x64\Lib\site-packages\nvidia\cusparselt\include'

@rwgk rwgk merged commit 7e26f1d into NVIDIA:main Oct 30, 2025
64 checks passed
@rwgk rwgk deleted the cusparselt_support branch October 30, 2025 20:44
@leofang leofang added this to the cuda.pathfinder 1.4.0 milestone Nov 10, 2025
@leofang leofang added the P0 High priority - Must do! label Nov 10, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cuda.pathfinder Everything related to the cuda.pathfinder module enhancement Any code-related improvements P0 High priority - Must do!

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Support cuSPARSELt

3 participants