Skip to content

Migrate size copies to vecmem async copies#1235

Merged
stephenswat merged 1 commit into
acts-project:mainfrom
stephenswat:cleanup/async_copy
Jan 20, 2026
Merged

Migrate size copies to vecmem async copies#1235
stephenswat merged 1 commit into
acts-project:mainfrom
stephenswat:cleanup/async_copy

Conversation

@stephenswat
Copy link
Copy Markdown
Member

This commit replaces the last remaining device-to-host copies into unpinned memory in our main tracking pipeline. It also replaces previous uses of traccc::cuda::get_size with the same mechanism.

@stephenswat stephenswat requested a review from krasznaa January 19, 2026 13:07
@stephenswat stephenswat added cleanup Makes the code all clean and tidy performance Performance-relevant changes labels Jan 19, 2026
@stephenswat
Copy link
Copy Markdown
Member Author

Peformance before at $\langle\mu\rangle=140$ (12 streams, RTX A5000):

Event processing  6.93284 ms/event, 144.241 events/s

And after:

Event processing  6.28273 ms/event, 159.166 events/s

@stephenswat
Copy link
Copy Markdown
Member Author

Closes #1223.

Copy link
Copy Markdown
Member

@krasznaa krasznaa left a comment

Choose a reason for hiding this comment

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

It will interfere a bit with my "seeding harmonization" branch, but we could go for it in the meanwhile.

I'll open a PR later today with that code of mine. Which gets rid of the seed_finding and spacepoint_binning sub-tools in all languages. So maybe we hold off on this for a bit, and then see how to best combine the two PRs? (I did not touch the CFK or the parameter estimation code yet.)

Comment thread device/cuda/src/seeding/seed_finding.cu Outdated
@stephenswat
Copy link
Copy Markdown
Member Author

It will interfere a bit with my "seeding harmonization" branch, but we could go for it in the meanwhile.

Removed the changes to files affected by your new PR.

@stephenswat
Copy link
Copy Markdown
Member Author

Physics performance summary

Here is a summary of the physics performance effects of this PR. Command used:

traccc_seeding_example_cuda --input-directory=/data/Acts/odd-simulations-20240506/geant4_ttbar_mu200 --digitization-file=geometries/odd/odd-digi-geometric-config.json --detector-file=geometries/odd/odd-detray_geometry_detray.json --grid-file=geometries/odd/odd-detray_surface_grids_detray.json --material-file=geometries/odd/odd-detray_material_detray.json --input-events=10 --use-acts-geom-source=on --check-performance --truth-finding-min-track-candidates=5 --truth-finding-min-pt=1.0 --truth-finding-min-z=-150 --truth-finding-max-z=150 --truth-finding-max-r=10 --seed-matching-ratio=0.99 --track-matching-ratio=0.5 --track-candidates-range=5:100 --seedfinder-vertex-range=-150:150

Seeding performance

Total number of seeds went from 298342 to 298342 (+0.0%)

Seeding plots



Track finding performance

Total number of found tracks went from 50191 to 50193 (+0.0%)

Finding plots









Track fitting performance

Fitting plots














Seeding to track finding relative performance

Seeding to track finding plots



Note

This is an automated message produced on the explicit request of a human being.

This commit replaces the last remaining device-to-host copies into
unpinned memory in our main tracking pipeline. It also replaces previous
uses of `traccc::cuda::get_size` with the same mechanism.
@sonarqubecloud
Copy link
Copy Markdown

Copy link
Copy Markdown
Member

@krasznaa krasznaa left a comment

Choose a reason for hiding this comment

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

👍

@stephenswat stephenswat enabled auto-merge January 20, 2026 10:39
@stephenswat stephenswat merged commit b1d5a22 into acts-project:main Jan 20, 2026
23 of 27 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cleanup Makes the code all clean and tidy performance Performance-relevant changes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants