1,159 Commits over 304 Days - 0.16cph!
Optim: avoid row allocations
Tests: ran unit tests
Optim: get rid of small Column allocs
- also removed all private qualifiers
Tests: ran unit tests
Update: set out optim plans
- reuse string builder via pool
- Renamed TextTableOriginal to TextTableNew (otherwise I'll accidentally break the game before I mean it)
Tests: none, trivial changes
Update: initial test setup to validate optimizations for TextTable
Tests: ran the new unit and perf tests
Merge: from eventrecord_allocs
- Reduces the number of allocations caused by our server-side analytics
- New "analytics.small_buffer_send_limit" persistent ServerVar to reduce task scheduling overhead. Set -1 to return original behavior.
Tests: ran existing analytics unit tests, booted server in editor.
Merge: from main
Tests: none, no conflicts
Clean: remove my testing setup
Test: none, trivial change
Optim: send small server-side analytics events using the same task thread
- Controlled via analytics.small_buffer_send_limit - to disable, set to -1, to enable for everything set to
999999
- Default to 16KB
- Preserved between server restarts
This avoids ~1KB of allocations just to schedule another async task per upload. On busy servers(100pop) this can save 0.8MB per frame.
Tests: booted in editor to check the command presence
Bugfix: EventRecord.AddField(bool) now respects it's param
Lucky for us, wasn't used anywhere outside of tests.
Tests: none, trivial change
Update: adding a couple perf tests for EventRecord
- also removed one of profilign scopes since I don't need it anymore
Used them to check if packing EventRecordField would give any perf benefit, and it's a no - indistinguishable from noise.
Tests: ran the new tests
LOD0 and prefabs setup condenser tanks smalls
Optim: Allocate scratch buffer on the stack instead of thread local mem
Perf tests showed same performance, so we can save on the global allocation
Tests: ran editor on craggy
Optim: avoid scratch buffer round trip
After checking internals, GUID serialization is also alloc-free, so routing through that.
Tests: profiled in editor
Optim: eliminate float/double related allocs in EventRecordField.Serialize
Need to run a couple additional experiments(stackalloc, tagged union), but this part is basically done.
Tests: ran in editor, observed in profiler that no more allocs are happening in Serialize for small records
Optim: EventRecordField - use thread local scratch to avoid GUID serialization allocs
Tests: validated value via debugging, unity profiler showed no allocs during Serialize(CSV) call
Update: hacky EventRecord profiling setup to track allocations
Will need to discard this before merge
Tests: ran in editor
Bugfix: Use valid index in WaterTestFromVolumes
Tests: detected during staging demo playback with useparallelupdatejobs - reran the demo, no more NREs
Merge: from parallel_validatemove
- Fixers a couple rare bugs leading to missing data from FullServerDemo recordings
- More work on BasePlayer.SErverUpdateParallel, still disabled
- Editor-only: Added a couple unit tests
- Editor-only: ServerDemoPlayer - disable error spam during demo playback, improve log format
- Editor-only: ServerDemoPlayer - automatically authenticate connections during demo playback
Tests: played back demo from staging server, recorded a couple new demos in local editor
Merge: from main
Tests: none, no conflicts
Bugfix: Initialize WaterSystem coarse grid on clients
- Also make it safe for scenes that don't have a water setup (like playground)
Tests: tested with a separate client connecting to server - saw that Client was initializing everything correctly in the right order
Bugfix: ensure WaterCollision grid is setup with right terrain dimensions
Tests: In editor, started on craggy - size matched. Started on procgen - matched. Played demo - matched. In all cases validated initialization order to confirm no colliders/volumes were added too early. Exported PNG of the grid.
Bugfix: FullServerDemos - transient entity recording fixes
- Use DemoCount instead of ChunkIndex when determing to send snapshots - ChunkIndices reset, so they can rarely overlap for an entity and cause it to not send a snapshot
- Reset counter on prefab pooling - previously it could cause skipping of transient entity snapshotting
Tests: recorded a bunch of demos in editor
Bugfix: ServerDemoPlayer - skip the auth flow when playing the demo
- also index packet logging messages - makes it easier to sort out order of things in editor logs
Seems the simpler thing to do for now.
Tests: played back staging demo - much less error spam
Update: ServerDemoPlayer - add Ready message logging
Tests: none, trivial change
Update: ServerDemoPlayer - log more message types
Tests: ran demo from staging server
Update: ServerDemoPlayer - switch to control error reporting
Tried the demo from staging, the error spam is too much and not super helpful, so adding a toggle to disable it.
Tests: Played back the staging demo
Optim: Avoid handling null cases in a batch with only non-null values
- Also updated the test to spawn entities, since now it's a requirement for the func.
This removes the need to juggle data to setup batch operations. Should save a bit of time.
Tests: Ran the updated unit tests
▄▇▅▆▇▊: ▇▆▆▊▉▍ ▄▋ ▊▉▆▇▅▉▋▇▉▄▅▋█▉▆ ▋▇▆▍█ ▅▊ ▅▅▋▌▅▅▄▍▄▆▍█▅▆▋█▇▉▆▋ ▌▅▄▊ ▅▌▍▉▆▇▍▊▉▄▇▄
▉▆▇▊▆: ▉▄▌▌▄▇ ▍▄▌▌ ▇ ▅▇▋▄ ▅▅▊▇ ▅▉▋▊▇▇▌▅▌▋▌▄▅▊▉▆▄▆▋ ▅█▄ █▇ █
Merge: from main
Tests: none, no conflicts
Merge: from profiling_improvements
- makes the Linux binaries compatible with more distros (Ubuntu 20.04, Debian 12)
Tests: ran in Ubunti 20.04 and 24.04 via WSL. Took snapshots and opened them in perfetto.
Update: ServerProfiler now usable on older Linux distros
- Binaries built using revision 45d79338
Should work on Debian 12.
Tests: Ran on WSL Ubuntu 20.04 and Ubuntu 24.04 and took snapshots - all worked.
Merge: from main
Tests: none, no conflicts
Merge: from profiling_improvements
- Fixes ProfilerBinViewer to display all available threads and fix invalid callstack depth calculation
- Fixes a bug that would prevent json from being generated on busy servers
- Fixes a bug with timelines being very-slightly out of sync
- Optim/Bugfix to filter out all constructors from being profiled
Tests: a lot of exports in the edittor and a bit of forced "bad" cases
Bugfix: ServerProfiler - properly filter out constructors
- Due to a typo (missing .) it would never match constructors and never filter them out
Using release libs built from ec8c5522
Tests: snapshot in editor on Craggy - confirmed no contructors were recorded.
Bugfix: ServerProfiler - sync up non-main-thread timelines to main thread
Previously it was possible to have a small gap at the very start of the snapshot on non-main thread views.
Tests: exported a snapshot in the editor.
Bugfix: ServerProfiler - properly handle sheared callstack at the start of worker threads during export
This revealed that I have a bug with non-main thread timing (things are slightly offset). Will fix next.
Tests: injected data that was tripping up the export originally. Validated that it does trip up before fix, and now exports correctly after fix. Exported 10 snapshots in the editor - no failures, and all look correct.
Bugfix: ProfileBinViewer - fix callstack depth calculation
- Also replace RadioBoxGroup for threads with a Dropdown - turns out 50 threads radio buttons don't vertically fit my monitor. Who knew.
Hoping it's the same issue in the json exporter.
Tests: Opened a debug snapshot from official server.
Merge: from main
Tests: none, no conflicts
Bugfix: CollectionUtil.SortInplace now correctly sorts
- Also renamed these utils as previous names were confusing
Tests: ran the old + new unit tests - all pass
Update: Adding unit tests for CollectionUtil
Discovered that my SortInplace is borked, so I'll fix it in the next update.
Tests: ran the new unit tests
Update: consolidate various scatter code used in batched WaterLevel and GamePhysics checks
Will cover them with tests just to be extra safe.
Tests: ran WaterLevel and GamePhysics unit tests
Update: Name a couple magic constants
Tests: none, trivial change
Clean: rogue newline
Tests: trivial change
Update: TerrainCollision.GetIgnore(Span...) now uses the ignore grid broadphase
Tests: ran new unit tests that go through this path
Update: new unit tests for GamePhysics.HandleIgnoreCollisions
- Also removed an unnecessary branch
Tests: ran the new unit tests
Merge: from main
Tests: none, no conflicts
▍▌█▇█: ▅▅▇▍ ▋▅▊▄▊█_▌▋▌▊█▌▊▋▌▉▋_▄▅▆▇▆▉▉
▄▍▅▇▅▌ ▆▇▇ ▆▌▉█▅▊ ▊▋▅▌▌▉ ▍▇▅▊▌▊▅
▍▅▊█▉: ▅▄▋▇ █▉▍█ ▅▉▇▌▆▄ ▉▇▉ ▉ ▅▇▋▍▄, ▊▍▊ ▇▅▄█ ▅▊▇▍▇▅▍ ▇▆▉▍▇█ ▇▍▆█▋▅ ▌▇▊█ ▅▌ ▅▄▄▇▌▆▌ █▊▇ ▊▉▇▋█▇.
█▉▋▅▍▆: ▊▌▌ ▇█▌▇▅▋▅ ▇▆▊▄▋▇▉██▄▅ ▉▍▉▌▇▅▍ ▉▄█ █▊▊ ▊▄▋▋▄▉▌▊
- ▉▉▍█▅ ▇▋▄▄▇ ▋▌ ▅▄▍▇█▅ ▋█▉ ▊▊▌▉▄█▊ ▊▌█▌▆ █▄ ▇▉█ ▅▋▆▉ ▄▅▋▌▉▉▅ █▆ ▌▌█▅▌ ▌▅▍▅▌▄▇▅ ▌▌▊ ▊▄▉▋, ▄▍▍▅▍▉▉ ▄▋ ▉▍ ▉▆▉▉▄▌ ▇▅█▇▉.
▄▍▅▅ ▌▆▌▅▌█ ▉▊▅▍ ▌▆▅▇▇▅ ▅▊▌█▅▊ ▇▌▌▆▍▄█ ▉▌▄▍ ▌▊▆▅▉▊ ▉▍▊▇▊▅▉▆▋▅ ▆▇█▄ ▋▄▆▍▄▇▊█ ▉▅▊▄ ▋▆ ▇▊▅+ ▍▉▍▇.
██▆▋▊: ▆▍▅▍▆▆ █▍▆▋ █▇▅▆ ▄▅██ - ▋▉█ ▆▍▅▍ ▆▋ █▋▆▊▄▆▇▅ ▊▆▉ ▍█▊▌ ▍▉▅▋▌. ▇▍▅ ▄▍█▅ ▆▍ ▍▄▍▍█▄▅ ▄▄ █▅▋ ▄▇▍▍▉▇▍ ▉▇█▄▅█ ▇█▉ ▆▊▉ ▄▅▅▊▆▊. ▍▍▊▊ ▍█ ▋ ▊▅▇▊▆▄▇▅ ▆▊▊▉▄▋ ▉▄▉ ▉▌ █▌▄▉▍▊▊▆▊ ▇▇▍ ▇▍▆ █ ▇▄▌▇ ▅▆▇▋ ▉▉ ▅▅▌▇▍▄▄▅▄▌ ▊▉▇▍▉▌.
Merge: from parallel_validatemove
* Modifies Full Server Demo recording to grab more data + timestamps per packet
* Adds editor-only "DemoServer" server backend that we can use in editor to play back full server demos. Switch editor to ServerMode and put path in GameSetup object
* New server-side batch-update of players routine - disabled by default as it's not validated yet. Controlled by `server.UsePlayerUpdateJobs`
* Added GamePhysics batch versions of OverlapCapsule and OverlapSphere
* Couple unit tests to check Water's batch queries against non-batched versions
Tests: A lot of server demos recorded and played back in the editor doing core activities (looting, harvesting, interacting). Made sure standalone server and client builds locally.
