url/colibri
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Profiling page: per-turn phase timings, live in the web dashboard

Browse Source 
The engine already tracks where each turn's wall time goes (expert-disk
service, I/O wait, expert matmul, attention, lm_head) — it just only spoke
at exit or under PROF=1. Stream it instead:

- glm.c: mux serve emits a per-turn "PROF" protocol line next to TIERS/HITS
  (window deltas per request, same convention as the STAT hit%); the phase
  window base is now always captured (a few loads per request).
- openai_server.py: parses PROF into a 120-turn rolling window and serves it
  at /profile (read-only, same trust level as /health).
- web: new Profiling tab — stat tiles (tok/s, wall, tokens/forward, disk
  service), wall-time composition bars for the last turn and the window,
  per-turn throughput and stacked phase columns with hover readouts, and a
  table of recent turns. Disk service is shown apart from the stack: it
  overlaps with compute, so only the I/O wait the compute thread felt counts
  inside wall time. Phase colours are a CVD-validated set with gaps + legend
  + table so identity never rides on colour alone.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WhTmF8yvEBgSkUKSVfZF7P
This commit is contained in:
Claude
2026-07-14 23:48:54 +00:00
committed by Nicholas Beerbower
parent 63a6824881
commit 6afffbcbf2
10 changed files with 321 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+3 -2
View File
@@ -540,9 +540,10 @@ What you get:
- **Chat** with live metrics: a flashing token counter while generating, then tok/s, time-to-first-token, prompt→completion counts and queue wait;
- **Runtime panel**: your hardware (CPU, GPUs + VRAM, RAM, cores), the scheduler, and the live expert-tier bar — how many of the 19,456 experts sit in VRAM / RAM / disk right now;
- **Brain**: the whole model as a 76×256 cortex, one cell per expert. Colour = tier, brightness = routing heat, and the experts routed in each turn flash white and decay — you watch the model think. Hover any cell for its tier, heat and [measured topic affinity](https://github.com/JustVugg/colibri/issues/175) (specialists for code, Chinese, math, law… live in layers 11–22).
- **Brain**: the whole model as a 76×256 cortex, one cell per expert. Colour = tier, brightness = routing heat, and the experts routed in each turn flash white and decay — you watch the model think. Hover any cell for its tier, heat and [measured topic affinity](https://github.com/JustVugg/colibri/issues/175) (specialists for code, Chinese, math, law… live in layers 11–22);
- **Profiling**: where each turn's wall time went — I/O wait vs expert matmul vs attention vs LM head — as stacked per-turn bars, plus throughput history, tokens-per-forward batching, and a table of the recent turns. The same phase timers behind the `PROFILE` line, streamed live.
The dashboard talks to the engine over two tiny protocol lines (`TIERS`, `EMAP`/`HITS`) and plain JSON endpoints — nothing heavier than the engine itself.
The dashboard talks to the engine over a few tiny protocol lines (`TIERS`, `EMAP`/`HITS`, `PROF`) and plain JSON endpoints — nothing heavier than the engine itself.
## Got a better machine? Try it — here's what to expect