Commit 75c6467
fix(replay): stamp frame points with the playhead date so series render as one line (#620)
ScatterGraph scopes Pareto frontiers and line paths per point.date (needed
for the comparison-date overlay), but replay frames copied each config's
template point verbatim — carrying that config's first-observation date.
Configs introduced on different dates therefore landed in different date
buckets within one frame, splitting a single hardware series into several
same-colored roofline segments with duplicated line labels ("one line
diverging into two") the moment the playhead crossed a newer config's first
date. Stamping every frame point with the playhead's current date restores
one frontier and one path per (hw, precision), which matches what a replay
frame is: a single point in time.
中文:replay(时间回放)图表中,同一硬件系列会分裂成多条同色曲线并出现重复的
曲线标签。原因是每帧数据沿用了各 config 模板点的首次观测日期,而 ScatterGraph
按 point.date 分桶计算 Pareto 前沿并分段绘制路径。现将每帧所有数据点统一标记为
播放头当前日期,使每个(硬件、精度)系列在单帧内只生成一条前沿曲线,符合
"一帧即一个时间点"的语义。附带回归单元测试。
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>1 parent 61128b4 commit 75c6467
2 files changed
Lines changed: 27 additions & 2 deletions
File tree
- packages/app/src/components/inference/replay
- __tests__
Lines changed: 20 additions & 1 deletion
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
17 | 17 | | |
18 | 18 | | |
19 | 19 | | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
20 | 23 | | |
21 | 24 | | |
22 | 25 | | |
| |||
38 | 41 | | |
39 | 42 | | |
40 | 43 | | |
41 | | - | |
| 44 | + | |
42 | 45 | | |
43 | 46 | | |
44 | 47 | | |
| |||
243 | 246 | | |
244 | 247 | | |
245 | 248 | | |
| 249 | + | |
| 250 | + | |
| 251 | + | |
| 252 | + | |
| 253 | + | |
| 254 | + | |
| 255 | + | |
| 256 | + | |
| 257 | + | |
| 258 | + | |
| 259 | + | |
| 260 | + | |
| 261 | + | |
| 262 | + | |
| 263 | + | |
| 264 | + | |
246 | 265 | | |
247 | 266 | | |
248 | 267 | | |
| |||
Lines changed: 7 additions & 1 deletion
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
5 | 5 | | |
6 | 6 | | |
7 | 7 | | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
8 | 14 | | |
9 | 15 | | |
10 | 16 | | |
11 | 17 | | |
12 | | - | |
| 18 | + | |
13 | 19 | | |
14 | 20 | | |
15 | 21 | | |
| |||
0 commit comments