bowenliang123 4136d0e565 fix(timing): open the decode window at the first marker, not the first token ...

A redacted reasoning block leaves no chunk behind, so the first token
lands after the reasoning window — pricing ttftMs/genMs off the token
charged the marker-tiled decode window to the wait as well, and the
legend's rows summed past 100%. The wait/generation boundary now sits
at the first observable instant (the earlier of the first token and the
first block marker), the same boundary the strip's spans already paint;
the TPS chip keeps the first-token anchor for harness parity.
stateVersion 23 → 24 refolds cached rows with the corrected totals.

2026-09-30 23:53:59 +08:00

..

client

feat(timing): the session-time strip — the session's active time as one packed band

2026-09-30 17:56:09 +08:00

host

fix(timing): open the decode window at the first marker, not the first token

2026-09-30 23:53:59 +08:00

shared

fix(timing): open the decode window at the first marker, not the first token

2026-09-30 23:53:59 +08:00