[GPU] Add optimized TopK kernel using radix histogram and bitonic sort.#34539
Open
hyunback wants to merge 1 commit intoopenvinotoolkit:masterfrom
Open
[GPU] Add optimized TopK kernel using radix histogram and bitonic sort.#34539hyunback wants to merge 1 commit intoopenvinotoolkit:masterfrom
hyunback wants to merge 1 commit intoopenvinotoolkit:masterfrom
Conversation
Add a new GPU TopK kernel (arg_max_min_topk_radix) that reduces complexity from O(N·K) to O(N + K·log²K) for f16 inputs with SORT_BY_VALUE mode. Two-level 256-bin radix histogram to find the K-th threshold in O(N), followed by SLM bitonic sort for O(K·log²K) final ordering Signed-off-by: hyunback <hyunback.kim@intel.com>
hyunback
added
WIP
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description of the issue(symptom, root-cause, how it was resolved)
Yolo26 shows poor performance, the main bottleneck is TopK.
Add a new GPU TopK kernel (arg_max_min_topk_radix) that reduces complexity from O(N·K) to O(N + K·log²K) for f16 inputs with SORT_BY_VALUE mode. Two-level 256-bin radix histogram to find the K-th threshold in O(N), followed by SLM bitonic sort for O(K·log²K) final ordering
Reproduction step and snapshot (if applicable. Do not attach for customer model)
$ ./benchmark_app -m ./yolo26n_int8_openvino_model/yolo26n.xml -d GPU -t 10
Checklist
Tickets: