Skip to content

GPU TPC: added dynamic buffer allocation during track-model decoding #14747

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
davidrohr merged 3 commits into from
Oct 18, 2025
Merged

GPU TPC: added dynamic buffer allocation during track-model decoding #14747

davidrohr merged 3 commits into from
Oct 18, 2025

Conversation

@cima22
Copy link
Contributor

@cima22 cima22 commented Oct 16, 2025

Dynamic buffer utilization for track-model decoding based on input data. Added margin based on the ratio between attached and unattached clusters.
Tests on PbPb simulated datasets, memSize limited to 15GB:

Interaction Rate Device Memory Before (Bytes) Device Memory After (Bytes)
5kHz 3,940,286,464 665,190,400
47kHz 5,647,106,048 3,087,728,640
50kHz 5,537,529,856 2,975,465,472

@cima22 cima22 requested a review from davidrohr as a code owner October 16, 2025 21:29
@github-actions
Copy link
Contributor

github-actions bot commented Oct 16, 2025

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5

@alibuild alibuild mentioned this pull request Oct 17, 2025
Closed
@cima22
Copy link
Contributor Author

cima22 commented Oct 17, 2025 •
edited
Loading

Tests with new implementation:

Dataset max Attached / SecRow max Unattached / SecRow Margin 1 Margin 2 Device memory (Bytes)
PbPb 5kHz 623 915 1372 1153 682,033,152
pp 200kHz 609 737 1105 1109 655,491,072
PbPb 47kHz 9207 10075 15112 20609 3,654,483,968
PbPb 50kHz 8849 9807 14710 19556 3,475,963,904

@cima22 cima22 force-pushed the dynamic_buffer_decompression branch from d8b4c8e to 0d68716 Compare October 17, 2025 13:49
@alibuild
Copy link
Collaborator

alibuild commented Oct 17, 2025

Error while checking build/O2/fullCI_slc9 for d8b4c8e at 2025-10-17 17:44:

## sw/BUILD/O2Physics-latest/log
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
[0 more errors; see full log]

Full log here.

@davidrohr davidrohr merged commit 95b425a into AliceO2Group:dev Oct 18, 2025
11 checks passed
plariono pushed a commit to plariono/AliceO2 that referenced this pull request Oct 23, 2025
@cima22 @plariono
GPU TPC: added dynamic buffer allocation during track-model decoding (A…
f8b9265 
…liceO2Group#14747)

* GPU TPC: added dynamic buffer allocation during track-model decoding

* GPU TPC: improvement for dynamic buffer size for track-model decoding

* GPU TPC: increased margins for track-model decoding buffers
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@davidrohr davidrohr Awaiting requested review from davidrohr davidrohr is a code owner

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@cima22 @alibuild @davidrohr