perf_census: the settings row says the census runs only in the import memory mode, since a mirror prints its own mem lines (#946)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Valiantsin DzerakhandClaude Opus 5.5 committed 2026-11-21 14:10:35 +03:00
1 parent a4c761c98c
commit ce47fdda3f
1 file changed
+1 -1
+1 -1
View File
@@ -32,7 +32,7 @@ Many rows below are developer tools that only act together with another setting
| `menu_cursor_log` | `1` logs what the menu cursor code reads at every screen change. | Only adds log lines. | Safe |
| `plugin_tick_offset` (default `0`) | Added to every GetSystemTick in the plugin service run (`totk.exe --plugin-init DIR`), so `tools/exefs_plugins.py` can run a plugin's init twice with other ticks and compare the records. | Has no effect outside `--plugin-init`; a plugin's init at the start of the game gets the plain ticks. | Safe |
| `plugin_trace` (default `0`) | `1` logs the first call of each entry of a mod plugin's hooks (its callbacks, replace and call targets) as `plugins: hook entry ... reached`. | Each such call goes through a counting step; nothing changes without a mod plugin. | Safe |
| `perf_census` (default `1`) | With `TOTK_TRACE=perf`, `0` turns off the dirty census: the `mem: dirty` lines that count, at every submit, the pages the CPU wrote in the pools the replay binds as buffers. | The census queries the write watch at every submit, 3.5-5 ms of the replay thread a frame on a Steam Deck; with `0` a perf trace measures the frame without that cost, and `tools/perf_summary.py` prints no dirty lines. | Safe |
| `perf_census` (default `1`) | With `TOTK_TRACE=perf` and the import memory mode (`TOTK_MEMORY=import`, the automatic mode on an integrated GPU; the mirror mode of a discrete GPU has its own `mem:` lines instead of the census), `0` turns off the dirty census: the `mem: dirty` lines that count, at every submit, the pages the CPU wrote in the pools the replay binds as buffers. | The census queries the write watch at every submit, 3.5-5 ms of the replay thread a frame on a Steam Deck; with `0` a perf trace measures the frame without that cost, and `tools/perf_summary.py` prints no dirty lines. | Safe |
| `perf_slow_ms` (default `50`) | Frames at least this many milliseconds long get a `perf: slow frame` log line with the time blocked per cause. | Changes how many frames are logged; a small value fills the log. | Safe |
| `replay_check` | `1` reads every texture read-back a second time after a full drain of the replay queue and compares the two. | Stalls the replay queue at every read-back, so frames get much slower; mismatches are only counted and logged. | Caution |
| `resolve_check` | `1` compares every cached pool lookup with the locked lookup and stops on a difference. | Slows every pool lookup; the game stops if the two ever differ. | Caution |