notes: Patch C2 Phase 7 — N=3 ramp, no measurable throughput delta #15

2026-05-08T05:43:57Z

claude-noether commented

2026-05-08 05:43:57 +00:00

N=3 reps on C2 (cleanups+F+G+D+E+C2 srcversion 619A51E61BF5479AAC146E6, fresh chip uptime 544/716/750 s):

rep	MB/s
1	2.289
2	2.165
3	2.376

N=3 mean 2.277 MB/s. vs Patch C v3 N=3 (2.352): -3% (within rep variance). vs Patch B baseline (1.362): +67%.

C2 was predicted in plan §4.5 as a possible <2% delta outcome — observed -3%, within noise. Ship for kernel.org-submission upstream-cleanliness anyway. The tasklet hop in ieee80211_rx_irqsafe was apparently cheap on this kernel.

Phase 8 lesson

Drop-in API correctness moves don't always pay perf dividends. Patch C v3 architectural change remains the durable +73% win; D / E / C2 / F / G are cleanup-class patches that don't visibly compound above noise.

Throughput ceiling on this hardware: ~2.4 MB/s sustained @ 4 MB/s sender, fresh chip. Further improvement requires firmware-side fixes (wsm_generic_confirm 0x0007), not driver-side.

Campaign closure

Patch C v3 + F + G + D + E + C2 series all merged + Phase-7-verified
Bug #5 RX-degradation hypothesis falsified (Phase 0 result)
Open work: Task #24 (post-cleanup observation, mostly answered) + Task #25 (Allwinner sw_mci_check_r1_ready measurement, low priority)

N=3 reps on C2 (cleanups+F+G+D+E+C2 srcversion `619A51E61BF5479AAC146E6`, fresh chip uptime 544/716/750 s): | rep | MB/s | |----:|----:| | 1 | 2.289 | | 2 | 2.165 | | 3 | 2.376 | N=3 mean **2.277 MB/s**. vs Patch C v3 N=3 (2.352): **-3%** (within rep variance). vs Patch B baseline (1.362): **+67%**. C2 was predicted in plan §4.5 as a possible <2% delta outcome — observed -3%, within noise. Ship for kernel.org-submission upstream-cleanliness anyway. The tasklet hop in `ieee80211_rx_irqsafe` was apparently cheap on this kernel. ## Phase 8 lesson Drop-in API correctness moves don't always pay perf dividends. Patch C v3 architectural change remains the durable +73% win; D / E / C2 / F / G are cleanup-class patches that don't visibly compound above noise. Throughput ceiling on this hardware: **~2.4 MB/s sustained @ 4 MB/s sender, fresh chip**. Further improvement requires firmware-side fixes (`wsm_generic_confirm 0x0007`), not driver-side. ## Campaign closure - Patch C v3 + F + G + D + E + C2 series all merged + Phase-7-verified - Bug #5 RX-degradation hypothesis falsified (Phase 0 result) - Open work: Task #24 (post-cleanup observation, mostly answered) + Task #25 (Allwinner sw_mci_check_r1_ready measurement, low priority)

claude-noether added 1 commit 2026-05-08 05:43:57 +00:00

notes: Patch C2 Phase 7 — N=3 ramp, no measurable throughput delta 02d3f4b222

| rep | uptime | MB/s |
|----:|-------:|-----:|
|   1 |   544s | 2.289|
|   2 |   716s | 2.165|
|   3 |   750s | 2.376|

N=3 mean: 2.277 MB/s.  vs Patch C v3 N=3 (2.352 MB/s): -3% (within
rep variance).  vs Patch B baseline (1.362 MB/s): +67%.

C2 was predicted in §4.5 of the Phase 4 plan as a possible
"<2% delta" outcome -> "ship for upstream-cleanliness anyway".
Observed -3% -> within noise -> ship.  The tasklet hop in
ieee80211_rx_irqsafe was apparently cheap on this kernel.

Phase 8 lesson: _irqsafe -> _rx_ni is a CORRECTNESS / kernel.org-
submission move, not a performance optimization.  Don't oversell
predicted throughput deltas without prior measurement.

Patch C v3 architectural win remains the durable +73%; D / E / C2 /
F / G are smaller cleanups that don't compound visibly above noise.

Throughput ceiling on this hardware: ~2.4 MB/s sustained @ 4 MB/s
sender, fresh chip.  Further improvement needs firmware-side fixes
(wsm_generic_confirm 0x0007 path), not driver-side.

marfrit merged commit 216c7c59b1 into main

2026-05-08 05:46:07 +00:00

marfrit referenced this issue from a commit

2026-05-08 05:46:08 +00:00

notes: Patch C2 Phase 7 — N=3 ramp, no measurable throughput delta (#15)

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: marfrit/besser#15