Repository navigation
Conversation
Contributor
Provide the Rust payload, group and scratch contracts, native CUB provider and Oxide integration with complete Block and Warp I/O. Signed-off-by: Trent Nelson <[email protected]>
Expose Block and Warp scans with explicit payload validity, prefix registration and native provider integration. Signed-off-by: Trent Nelson <[email protected]>
Avoid copying inactive batch output lanes while preserving active results and add focused coverage for narrow output batches. Signed-off-by: Trent Nelson <[email protected]>
Provide scalar and payload reductions with validity-aware results. Include checked stateless and uniform i32 capture callbacks at their first public consumer, together with the native callback ABI. Signed-off-by: Trent Nelson <[email protected]>
Signed-off-by: Trent Nelson <[email protected]>
Preserve the batch dimension and per-batch output validity while using the CUB output-copy guard. Include the provider shape correction. Signed-off-by: Trent Nelson <[email protected]>
Expose complete Block and Warp exchange operations and Block shifts with explicit layout, output validity and scratch completion. Signed-off-by: Trent Nelson <[email protected]>
Provide neighboring-item transforms and head/tail flags with typed neighbor controls, payload validity and explicit scratch completion. Signed-off-by: Trent Nelson <[email protected]>
Expose Block histogram algorithms with typed bin controls and fresh output ownership through the shared native provider. Signed-off-by: Trent Nelson <[email protected]>
Group complete radix key, pair and rank operations with their validity, layout, bit-range and scratch contracts. Signed-off-by: Trent Nelson <[email protected]>
Provide Block and Warp key and pair sorting with partial inputs. Include the supported stateless full-input i32 Block comparator at its owning family and retain the checked callback ABI. Signed-off-by: Trent Nelson <[email protected]>
Provide minimum and maximum key and pair selection with full and partial input contracts, typed controls and explicit completion. Signed-off-by: Trent Nelson <[email protected]>
Provide windowed and output-directed decoding with typed run lengths, fresh decoded payloads and explicit scratch completion. Signed-off-by: Trent Nelson <[email protected]>
tpn
force-pushed
the
codex/cuda-coop-rust-run-length-decode
branch
from
October 25, 2026 01:33
4c5794c to
9d5615e
Compare
Contributor
|
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. |
tpn
changed the title
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Add Run Length Decode windows and checked destination writes. Typed run lengths, fresh decoded payloads and scratch completion preserve the CUB operation contracts.
Builds on #12246. The owning comparison isolates this review unit; the GitHub diff against
mainis cumulative.Validation on the reconstructed layer: formatting, host tests, clippy and docs, native provider coverage and standalone public examples on CUDA 13.3 / sm_120. Historical broader diagnostics, sanitizer and profile evidence retains its original source attribution.
This PR remains draft.