Skip to content

Conversation

srinivasyadav18
Copy link
Contributor

@srinivasyadav18 srinivasyadav18 commented Sep 3, 2025

Description

closes #5657

TODO

  • Move device_memory_resource to cuda/std/__device_memory_resource.h
  • Add test catch2_device_scan_env_api.cu
  • Add test catch2_device_scan_env.cu
  • Fix documentation

Copy link
Contributor

copy-pr-bot bot commented Sep 3, 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.

@cccl-authenticator-app cccl-authenticator-app bot moved this from Todo to In Progress in CCCL Sep 3, 2025
@srinivasyadav18 srinivasyadav18 marked this pull request as ready for review September 4, 2025 23:10
@srinivasyadav18 srinivasyadav18 requested a review from a team as a code owner September 4, 2025 23:10
@cccl-authenticator-app cccl-authenticator-app bot moved this from In Progress to In Review in CCCL Sep 4, 2025
Copy link
Contributor

github-actions bot commented Sep 4, 2025

📖 Doc Preview CI

🚀 Preview URL: https://NVIDIA.github.io/cccl/pr-preview/pr-5767/

Preview will be available once GitHub Pages deployment completes.

@gonidelis gonidelis self-requested a review September 5, 2025 01:53

This comment has been minimized.

@gonidelis
Copy link
Member

I guess this PR fixes this #5606?

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

@github-project-automation github-project-automation bot moved this from In Review to In Progress in CCCL Sep 19, 2025
// Equivalent to `cuexec::require(cuexec::determinism::run_to_run)` and
// `cuexec::require(cuexec::determinism::not_guaranteed)`
auto env = stdexec::env{cuda::execution::require(determinism_t{}), // determinism
allowed_kernels(kernels), // allowed kernels for the given determinism
Copy link
Contributor

Choose a reason for hiding this comment

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

could not be dangerous with multiple translation units?

Copy link
Member

@gonidelis gonidelis Sep 23, 2025

Choose a reason for hiding this comment

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

@fbusato could you please expand on it? I don't fully understand what the problem might be.
@srinivasyadav18 Is not_guaranteed in the set of deterministic requirements supported by ExlusiveScan? I don't see it anywhere in the header source and/or the underlying implementation.

Copy link
Contributor

@fbusato fbusato Sep 23, 2025

Choose a reason for hiding this comment

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

potential problem (I could be wrong): kernel functions are passed with pointers. Translation units could have difference pointer values for the same instantiation.

@github-project-automation github-project-automation bot moved this from In Progress to In Review in CCCL Sep 22, 2025

This comment has been minimized.

srinivasyadav18 and others added 2 commits September 23, 2025 12:36
- add [[no_discard]] attribute to New DeviceScan env based overload's
- reject not_guaranteed determinism with static_assert, as its not implemented
- store and check error in tests
- remove ifconstexpr in kernel tests

This comment has been minimized.

Copy link
Contributor

🥳 CI Workflow Results

🟩 Finished in 6h 32m: Pass: 100%/185 | Total: 6d 03h | Max: 3h 38m | Hits: 76%/186975

See results here.

@srinivasyadav18 srinivasyadav18 merged commit 843500e into NVIDIA:main Sep 25, 2025
194 of 195 checks passed
@github-project-automation github-project-automation bot moved this from In Review to Done in CCCL Sep 25, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Status: Done
Development

Successfully merging this pull request may close these issues.

Add env based overloads for cub::DeviceScan Algorithms
4 participants