EdnunpandClaude Opus 5
dfdc243f03
Two bugs behind the 2026-08-14 field report ("the latency slider does nothing" — a WASAPI-only
machine, auto-tune off, a clap test showing no change either way, while Ed's own two-slider rig
audibly works). Both found by MEASURING, not reasoning: --latency-lab drives the real playout
with realtime-paced sender/device shapes and records actual buffered depth once a second.
1. The slider wrote to a value nothing read. ReconcileReplicasLocked tags every session with an
OUTPUT LANE (WasapiLane/AsioLane — never Mixed while a device is ticked), and the render path
resolved its target from that lane. But MainForm.MaxLatencyBoxRoute sends the single slider to
RenderRoute.Mixed in EVERY non-BothIndependent mode. So in classic modes the slider — and the
auto-tune with it — updated a value the audio never consumed, leaving the real target on
LaneLatency's 30 ms default for the whole session; raise AND lower equally inert (lower also
matched on route, so it skipped every playing session). Only BothIndependent worked, because
there the slider writes WasapiLane, which IS what its sessions read. That's precisely "works on
mine, dead on his". Fix: the lane tag says which DEVICE a session renders through, not which
knob governs it. One slider now means ONE value (sharedLatency) read by every session whatever
its lane; two sliders only in BothIndependent, seeded from the shared value on entry so audio
doesn't jump. Arming also uses the session's own route target (matters in BothIndependent).
Measured, classic path: -0.15ms/s (STALLED) -> +2.07ms/s (GROWS).
2. Even correctly wired, a raise crawled. Steady-state depth feedback is capped at 0.3% (~3ms of
catch-up per second) AND only recomputed at the 10s drift-window boundary, so a 400 ms raise
took over two minutes and still read as a dead slider (Ed: "it needs to change the value
quickly not over minutes"). A raise can't be met by dropping audio, only by banking it, so a
DELIBERATE move (the hard setter — never the auto-tune's soft one, leaving its parked descent
behaviour untouched) now engages a fast approach: 5% bias, recomputed 5x/sec, self-clearing
within 15 ms of target. Audible glide, no gap, no click — the user asked for the change and
hears it happen. Measured: 49ms -> 324ms within 10s, then holds.
Gate: new step pins the invariant fast and deterministically — what the slider sets IS what the
session's route reads, a lower reaches the session, and two-slider mode keeps its lanes separate.
74/74 + relay. --latency-lab (six scenarios) / --latency-lab classic (the faithful app path) stay
in the tree as the measuring harness.
NOT released. Test build for the reporting user: D:\Dropbox\remsound-latency-slider-test.zip
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
114 lines
5.8 KiB
Plaintext
114 lines
5.8 KiB
Plaintext
RemSound latency lab - measuring depth-target convergence with the shipped playout code.
|
|
Raise under test: 30ms -> 330ms. Expected growth at full depth-bias: ~3ms/s.
|
|
|
|
--- T1 clean: write 10ms/10ms, read 10ms/10ms ---
|
|
settled at target 30ms: buffered=38ms ratio=1.001000 updates=1
|
|
t+ 5s buffered= 10ms ratio=1.000101 updates=2
|
|
t+10s buffered= 34ms ratio=1.000101 updates=2
|
|
t+15s buffered= 49ms ratio=1.000070 updates=3
|
|
t+20s buffered= 74ms ratio=1.000070 updates=3
|
|
t+25s buffered= 78ms ratio=1.000049 updates=4
|
|
t+30s buffered= 103ms ratio=1.000049 updates=4
|
|
t+35s buffered= 98ms ratio=1.000635 updates=5
|
|
t+40s buffered= 140ms ratio=1.000635 updates=5
|
|
t+45s buffered= 132ms ratio=0.999845 updates=6
|
|
t+50s buffered= 148ms ratio=0.999845 updates=6
|
|
t+55s buffered= 184ms ratio=0.999892 updates=7
|
|
t+60s buffered= 200ms ratio=0.999892 updates=7
|
|
RAISE result: 38ms -> 200ms in 60s = 2.70ms/s GROWS (mechanism working)
|
|
LOWER result: back to target 30ms -> buffered=69ms after 3s
|
|
|
|
--- T1 opus-frame: write 20ms/20ms, read 10ms/10ms ---
|
|
settled at target 30ms: buffered=76ms ratio=0.998000 updates=1
|
|
t+ 5s buffered= 58ms ratio=0.999200 updates=2
|
|
t+10s buffered= 97ms ratio=0.999200 updates=2
|
|
t+15s buffered= 126ms ratio=0.998841 updates=3
|
|
t+20s buffered= 127ms ratio=0.998841 updates=3
|
|
t+25s buffered= 158ms ratio=1.000088 updates=4
|
|
t+30s buffered= 153ms ratio=1.000088 updates=4
|
|
t+35s buffered= 177ms ratio=1.000062 updates=5
|
|
t+40s buffered= 202ms ratio=1.000062 updates=5
|
|
t+45s buffered= 207ms ratio=0.999144 updates=6
|
|
t+50s buffered= 237ms ratio=0.999144 updates=6
|
|
t+55s buffered= 266ms ratio=1.000001 updates=7
|
|
t+60s buffered= 271ms ratio=1.000001 updates=7
|
|
RAISE result: 76ms -> 271ms in 60s = 3.25ms/s GROWS (mechanism working)
|
|
LOWER result: back to target 30ms -> buffered=59ms after 3s
|
|
|
|
--- T1 gulpy device: write 20ms/20ms, read 21ms + 42ms gulp each ~1s ---
|
|
settled at target 30ms: buffered=43ms ratio=1.000400 updates=1
|
|
t+ 5s buffered= 41ms ratio=0.999771 updates=2
|
|
t+10s buffered= 59ms ratio=0.999771 updates=2
|
|
t+15s buffered= 53ms ratio=1.000529 updates=3
|
|
t+20s buffered= 85ms ratio=1.000529 updates=3
|
|
t+25s buffered= 97ms ratio=1.000460 updates=4
|
|
t+30s buffered= 130ms ratio=1.000460 updates=4
|
|
t+35s buffered= 142ms ratio=0.999214 updates=5
|
|
t+40s buffered= 140ms ratio=0.999214 updates=5
|
|
t+45s buffered= 158ms ratio=1.000737 updates=6
|
|
t+50s buffered= 189ms ratio=1.000737 updates=6
|
|
t+55s buffered= 180ms ratio=0.999408 updates=7
|
|
t+60s buffered= 238ms ratio=0.999408 updates=7
|
|
RAISE result: 43ms -> 238ms in 60s = 3.25ms/s GROWS (mechanism working)
|
|
LOWER result: back to target 30ms -> buffered=59ms after 3s
|
|
|
|
--- T1 jittery net: write 20ms +/-15ms, read 10ms/10ms ---
|
|
settled at target 30ms: buffered=70ms ratio=1.002000 updates=1
|
|
t+ 5s buffered= 50ms ratio=1.001100 updates=2
|
|
t+10s buffered= 39ms ratio=1.001100 updates=2
|
|
t+15s buffered= 79ms ratio=1.001370 updates=3
|
|
t+20s buffered= 47ms ratio=1.001370 updates=3
|
|
t+25s buffered= 85ms ratio=1.000359 updates=4
|
|
t+30s buffered= 89ms ratio=1.000359 updates=4
|
|
t+35s buffered= 82ms ratio=0.999952 updates=5
|
|
t+40s buffered= 117ms ratio=0.999952 updates=5
|
|
t+45s buffered= 132ms ratio=1.001166 updates=6
|
|
t+50s buffered= 162ms ratio=1.001166 updates=6
|
|
t+55s buffered= 162ms ratio=0.999619 updates=7
|
|
t+60s buffered= 189ms ratio=0.999619 updates=7
|
|
RAISE result: 70ms -> 189ms in 60s = 1.98ms/s SLOW (partially working)
|
|
LOWER result: back to target 30ms -> buffered=69ms after 3s
|
|
|
|
--- T2 engine-Mixed clean: write 20ms/20ms, read 10ms/10ms ---
|
|
settled at target 30ms: buffered=74ms ratio=1.001001 updates=1
|
|
t+ 5s buffered= 34ms ratio=1.000401 updates=2
|
|
t+10s buffered= 21ms ratio=1.000401 updates=2
|
|
t+15s buffered= 29ms ratio=1.000281 updates=3
|
|
t+20s buffered= 29ms ratio=1.000281 updates=3
|
|
t+25s buffered= 30ms ratio=0.999897 updates=4
|
|
t+30s buffered= 34ms ratio=0.999897 updates=4
|
|
t+35s buffered= 37ms ratio=0.999928 updates=5
|
|
t+40s buffered= 38ms ratio=0.999928 updates=5
|
|
t+45s buffered= 29ms ratio=0.999949 updates=6
|
|
t+50s buffered= 30ms ratio=0.999949 updates=6
|
|
t+55s buffered= 39ms ratio=1.000864 updates=7
|
|
t+60s buffered= 9ms ratio=1.000864 updates=7
|
|
RAISE result: 74ms -> 9ms in 60s = -1.08ms/s STALLED (the field bug)
|
|
LOWER result: back to target 30ms -> buffered=29ms after 3s
|
|
|
|
--- T2 engine-Mixed gulpy: write 20ms/20ms, read 21ms + 42ms gulp ---
|
|
settled at target 30ms: buffered=59ms ratio=0.998399 updates=1
|
|
t+ 5s buffered= 56ms ratio=0.999568 updates=2
|
|
t+10s buffered= 28ms ratio=0.999568 updates=2
|
|
t+15s buffered= 25ms ratio=0.999818 updates=3
|
|
t+20s buffered= 50ms ratio=0.999818 updates=3
|
|
t+25s buffered= 44ms ratio=0.999962 updates=4
|
|
t+30s buffered= 48ms ratio=0.999962 updates=4
|
|
t+35s buffered= 39ms ratio=0.999465 updates=5
|
|
t+40s buffered= 20ms ratio=0.999465 updates=5
|
|
t+45s buffered= 35ms ratio=0.999116 updates=6
|
|
t+50s buffered= 50ms ratio=0.999116 updates=6
|
|
t+55s buffered= 53ms ratio=0.999471 updates=7
|
|
t+60s buffered= 50ms ratio=0.999471 updates=7
|
|
RAISE result: 59ms -> 50ms in 60s = -0.15ms/s STALLED (the field bug)
|
|
LOWER result: back to target 30ms -> buffered=62ms after 3s
|
|
|
|
|
|
=== SUMMARY ===
|
|
T1 clean: write 10ms/10ms, read 10ms/10ms: raise 2.70ms/s [GROWS (mechanism working)], lower -> 69ms
|
|
T1 opus-frame: write 20ms/20ms, read 10ms/10ms: raise 3.25ms/s [GROWS (mechanism working)], lower -> 59ms
|
|
T1 gulpy device: write 20ms/20ms, read 21ms + 42ms gulp each ~1s: raise 3.25ms/s [GROWS (mechanism working)], lower -> 59ms
|
|
T1 jittery net: write 20ms +/-15ms, read 10ms/10ms: raise 1.98ms/s [SLOW (partially working)], lower -> 69ms
|
|
T2 engine-Mixed clean: write 20ms/20ms, read 10ms/10ms: raise -1.08ms/s [STALLED (the field bug)], lower -> 29ms
|
|
T2 engine-Mixed gulpy: write 20ms/20ms, read 21ms + 42ms gulp: raise -0.15ms/s [STALLED (the field bug)], lower -> 62ms
|