Fr0zkaandClaude Opus 5.5 b370908174 docs(assembler): round L10 report -- spawn waits on the full shape; 24 m crossing hits bad_alloc ...

First tile starts ~300 ms after cold start: plan 51 ms (serial), shape compile 223 ms (caves,
signs, detail band, far parents). The first SurfaceWorld tiles use the native heightfield and do
not read that shape, but generateTile requires the full base shape. 8/16 m crossings repeat no
compile (61 ms for 288 tiles); a 24 m crossing died with std::bad_alloc under the 6 GiB
address-space cap (cause unknown). Telemetry not committed (+1% when off).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CSFYgsdW8ejZSyiMf4qRSX

2026-10-19 23:44:52 +02:00

..

Linux1

wip(assembler): round L1 -- Linux baseline on abyss: Linux world identical to Windows

2026-10-19 18:18:34 +02:00

Linux2

wip(assembler): round L2 -- the cross-platform hash check can fail; FP contraction changes the world

2026-10-19 18:18:34 +02:00

Linux3

wip(assembler): round L3 -- Core forces FP contraction off and refuses to generate if it can't

2026-10-19 18:51:07 +02:00

Linux4

wip(assembler): round L4 -- lazy per-thread caches: 17.3 MiB -> 43 KiB per thread, view 7% faster

2026-10-19 19:16:14 +02:00

Linux5

wip(assembler): rounds L5+L6 -- Linux far-terrain captures and screen-space gate (world unchanged)

2026-10-19 21:10:08 +02:00

Linux6

wip(assembler): rounds L5+L6 -- Linux far-terrain captures and screen-space gate (world unchanged)

2026-10-19 21:10:08 +02:00

Linux7

wip(assembler): round L7 -- shared column cache across workers: whole view 16-19% faster, world unchanged

2026-10-19 21:46:15 +02:00

Linux8

docs(assembler): round L8 report -- where the time goes after the shared cache

2026-10-19 22:10:05 +02:00

Linux9

wip(assembler): round L9 -- LOD1 skips proven-uniform blocks: 95% fewer samples, view ~2% faster

2026-10-19 23:14:16 +02:00

Linux10

docs(assembler): round L10 report -- spawn waits on the full shape; 24 m crossing hits bad_alloc

2026-10-19 23:44:52 +02:00

Round1

wip(assembler): M3 round 2 -- islands get real mass: deep crust, separate tapered roots, arched spans

2026-10-10 12:24:35 +02:00

Round2

wip(assembler): M3 round 2 -- islands get real mass: deep crust, separate tapered roots, arched spans

2026-10-10 12:24:35 +02:00

Round3

wip(assembler): M3 round 3 -- separated islands + Unreal island route (owner build 18:46)

2026-10-10 20:53:03 +02:00

Round4

wip(assembler): M3 round 4 -- plans built inside Unreal from the owner's strates; no import needed

2026-10-10 22:48:46 +02:00

Round5

wip(assembler): round S1 -- one Core voxel-size source; per-tile feature lists (islands)

2026-10-18 18:48:38 +02:00

Round6

wip(assembler): round S2 (Sol) -- shared per-tile feature queries for every Core family; island placement 30x faster

2026-10-18 19:50:45 +02:00

Round7

wip(assembler): round S3 -- far ground from the native heightfield; LOD7/8 tiles 36-66% cheaper

2026-10-18 21:29:52 +02:00

Round8

wip(assembler): round S4 -- continuous far ceiling at the strate's upper boundary

2026-10-19 00:41:13 +02:00

Round9

wip(assembler): round S5 -- streaming gates + trace; coarse-first order kept opt-in (no win)

2026-10-19 04:42:41 +02:00

Round10

wip(assembler): round S6 -- in-game cold view counted; workers mostly idle, apply cap not the floor

2026-10-19 06:46:54 +02:00

Round11

wip(assembler): round S7 -- workers kept busy: in-game cold full view 17.5 s -> 7.8 s

2026-10-19 08:10:10 +02:00

Round12

wip(assembler): round S8 -- region rebuilds no longer stall admission; result backlog capped

2026-10-19 09:43:03 +02:00

Round13

wip(assembler): round S9 -- collision look-ahead: ground ring ready on arrival up to 40 m/s and 50 m/s falls

2026-10-19 10:56:54 +02:00

Round14

wip(assembler): round S10 -- LOD8 horizon sampled at MC spacing: LOD8 worker time 18.8 s -> 4.3 s

2026-10-19 12:23:18 +02:00

Round15

wip(assembler): round S11 -- bounded spawn apply budget: in-game full set 7.1 s -> 6.0 s

2026-10-19 13:03:44 +02:00

Round16

wip(assembler): round S12 -- short ready queue keeps workers fed between ticks: full set 6.0 s -> 4.9 s

2026-10-19 14:29:38 +02:00

Round17

wip(assembler): round S13 -- tile cost counted per LOD (diagnostics only, no output change)

2026-10-19 15:40:25 +02:00

Round18

wip(assembler): round S14 -- 4-way column cache + memoized MC gradients (output unchanged)

2026-10-19 17:58:54 +02:00

HANDOFF-ABYSS.md

docs(assembler): handoff for the abyss (Linux) research machine

2026-10-19 16:11:06 +02:00