SPB Git

Repository Public

PERFORMANCE.md 3.4 KB 49 lines markdown

1# NEXTERM — Performance (measured, not estimated)23**Method.** `Scripts/perf.sh` drives the Release build through System Events in an isolated data4directory: launch time is measured from `open` until the first on-screen window (CGWindowList probe5every 20 ms); memory is the process RSS from `ps`; CPU is `ps %cpu` (a *cumulative* average since6process start, so values taken right after creating many panes overstate idle load — see notes);7throughput pipes 200 MB of text through one pane with `yes | head -c` and times it in the shell;8restoration quits (which saves) and relaunches a 32-pane workspace.910**Environment.** Apple M5 Max (18 cores), 48 GB, macOS 27.0 (26A5425a), Xcode 27.2, Release11configuration, SwiftTerm Metal renderer, Menlo 13 pt, external 5K display as main screen.12Numbers are from a single run on 2026-11-01; re-run `Scripts/perf.sh` to refresh.1314## Results15| Metric | Value | Notes |16|---|---|---|17| Hardware | Apple M5 Max, 18 cores, 48 GB | macOS 27.0 |18| Build | Release configuration | NEXTERM 0.1.0 |19| Launch → first window | 350 ms | fresh workspace, 1 shell |20| Memory, 1 session | 115 MB RSS | after 2 s |21| Idle CPU, 1 session | 0,0 % | ps %cpu after 5 s idle |22| Memory, 4 sessions | 117 MB RSS | 4 child shells alive; idle CPU 0,0 % |23| Memory, 8 sessions | 119 MB RSS | 8 child shells alive; idle CPU 0,0 % |24| Memory, 16 sessions | 124 MB RSS | 16 child shells alive; idle CPU 0,2 % |25| Memory, 32 sessions | 132 MB RSS | 32 child shells alive; idle CPU 0,1 % |26| Sustained output | 40 MB/s | 200 MB via `yes | head -c`, 32 panes open, CPU during burst 4,9 % |27| Idle CPU, 32 sessions | 0,1 % | after burst settled |28| Restore 32-pane workspace | window 594 ms, all 32 shells 648 ms | layout restored, new processes |29| Memory after restore | 130 MB RSS | 32 sessions |3031## Notes and limitations32* **Launch 350 ms** (open → first on-screen window) includes creating the SQLite database, starting CLI33  discovery (async) and spawning the first login shell; binary already in the file cache. A first run after34  a reboot will be slower.35* **Per-pane memory** ≈ (132 − 115) / 31 ≈ **0.55 MB RSS per additional pane** at the app level; each pane36  also owns a zsh process (not counted). SwiftTerm allocates scrollback lazily (10 000 lines configured).37* **Idle CPU** stays at 0–0.2 % with 1–32 sessions (cumulative `ps %cpu`, so this includes the split storm).38* **Sustained output 40 MB/s** for 200 MB of text into one maximized pane with 31 other panes alive, app CPU39  4.9 % cumulative during the burst — the UI stayed responsive (System Events kept driving menus).40* **Restoration** of a 32-pane workspace: first window at 594 ms, all 32 new shells running at 648 ms.41* **Input latency** was not instrumented (no automated keystroke→frame probe exists yet); interactive42  use (vim, Claude Code TUI) showed no perceptible lag during the acceptance runs.43* **Resize storms**: live window resizing with vim/top open was exercised manually during acceptance44  tests (grid updates followed the window, no stale frames after the container-sizing fix).45* Memory leaks were not profiled with Instruments in this session (Instruments' CoreDevice plug-in is46  broken on this Xcode/macOS pair); the session registry removes terminal views from their superview on47  close and all observers are owned by the container view.48* Not measured: Intel, multi-GPU, or App Nap behaviour in the background.49
Files main
Menu

Theme