Improve multi-threaded performance of TGeoNavigator - #24
Open
sawenzel wants to merge 3 commits into
Open
Conversation
added 3 commits
July 28, 2026 10:13
This commit improves multithreaded TGeo navigation. The changes come from profiling a parallel geometry scan in ALICE. AddNavigator() inserts into fNavigators under fgMutex, but GetCurrentNavigator(), GetListOfNavigators() and SetCurrentNavigator() called find() on the same std::map with no lock. A thread booking its navigator rebalances the tree underneath a concurrent reader, returning a garbage TGeoNavigatorArray* or crashing outright. The lookup is factored into FindNavigatorArray(), which takes fgMutex in MT mode and skips it otherwise, matching what AddNavigator() already does. The TLS cache in GetCurrentNavigator() is untouched, so steady-state navigation stays lock-free. Assisted-by: Claude Code (review, hardening and benchmarking) Supervised-by: Sandro Wenzel <sandro.wenzel@cern.ch>
This commit improves multithreaded TGeo navigation. The changes come from profiling
a parallel geometry scan in ALICE: filling the material budget LUT on 28 cores goes
from 12x to 23x speedup (139 s -> 72 s).
Two costs dominate. Every per-thread scratch lookup went through TGeoManager::ThreadId(),
a non-inlined cross-library __tls_get_addr call, paid on every boolean, section or
division query. And the per-object ThreadData_t blocks are small allocations made back to
back, so slots of different threads share a cache line and invalidate each other on every
write -- false sharing, worst across sockets.
Each object now gets a dense index and the per-thread state lives in one thread_local
vector indexed by it. GetThreadData() becomes a header-inline TLS read plus an indexed
load, and each thread owns its whole vector.
TGeoBoolNode, TGeoVolumeAssembly, TGeoPatternFinder, TGeoPgon, TGeoXtru
TGeoPgon/TGeoXtru: noexcept move ctor on ThreadData_t (owns heap buffers, and for Xtru
a TGeoPolygon aliasing them) so a slot survives vector growth
TGeoPatternFinder: matrix reused across generations -- it is owned by the geometry
manager and never released, so a fresh one would leak per (thread, finder)
Provisioning goes away: ClearThreadData() bumps a generation counter and each thread
rebuilds its slot lazily, so CreateThreadData()/SetMaxThreads() are no longer needed to
make a given thread count work -- any thread count works out of the box. The counters are
atomic because ClearThreadData() is const and can run concurrently.
Assisted-by: Claude Code (review, hardening and benchmarking)
Supervised-by: Sandro Wenzel <sandro.wenzel@cern.ch>
This commit improves multithreaded TGeo navigation. The changes come from profiling a parallel geometry scan in ALICE. Eight threads navigating the same geometry must give exactly the single-threaded answer. Covers TGeoXtru, TGeoPgon, TGeoVolumeAssembly, TGeoBoolNode (composite shape) and TGeoPatternFinder (divided volume). The threads book their navigators lazily, so it also exercises AddNavigator() against the navigator-map readers. Assisted-by: Claude Code (review, hardening and benchmarking) Supervised-by: Sandro Wenzel <sandro.wenzel@cern.ch>
Test Results0 tests 0 ✅ 0s ⏱️ Results for commit 5fe298d. |
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.
Needed for AliceO2Group/AliceO2#15641