igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

class v5 design page: Devnet 3's crossing at 68,400 (epoch 19, program 3d375a55029e7e60 at attempt 2), the worker fault (generator-5 packs refused by the pre-kit workers) and its fix

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-08 11:56:24 +00:00
parent cabd38c01a
commit 4ed9be047a
1 changed files with 1 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
docs/design/class-v5-stored-state.md
View file

@ -46,6 +46,7 @@ Clock rule for this page: every time is UK time (UK time tonight, UTC+1). The bu
| 05:2x | F1 PASS on the freeze (100,000 class v5 programs, max saved share 4.688 percent, 0 over 5, 0 mismatches, 0 panics): the attack-pass families on class v5 read F8 PASS with the known residue, F4 PASS (AP-F4-1 fixed and passed), F9 PASS, F1 PASS |
| 09:4x | The Arc B580 read (09:36 UK): no fingerprint, the kit worker's self-test fails on both classes' vector lanes with the dataset right: an Intel OpenCL per-nonce evaluation fault, not class v5's; the Intel kit HELD out of 0.3.24 with the crossing time 09:36 UK; the kit stands on five of six platforms |
| 12:3x | 0.3.25's pairing line sent at the 12:30 default: the freeze 1c420786, kit zip 65b47211 (packs, ids and fingerprint byte-identical to the freeze's; built from 8f481459 with the Intel rotate-fold fix, gate 73 GREEN, suite green), the Intel worker held out; the second PC dark since 10:46 UK, so the Arc read and the Intel kit go to 0.3.26 at the earliest |
| 13:1x | Devnet 3 crossed the class v5 floor at DAA 68,400 (12:48 UK, by the floor: the v5 signal share read 1,407 bps). The template read on the seed at 13:1x UK: epoch 19, class 5, version 1538 (object byte 6), seed a75c5624...9d19, era the genesis, day 20,734; the executor serves epoch 19's stream (869 records, root 0x1fd55139...4561); the kit's igneum-pow names the program: attempt 2, id 3d375a55029e7e60. No block came for a time: every fleet miner's worker was the hive package's CUDA worker, which runs generators 2, 3 and 4 only and refuses the pack with a line ("program pack generator 5 is not a generator version this worker runs"), the miner at 0 MH/s retrying; the DMG's Metal worker from the freeze's tree refuses the same way (the v5 path is the v5-kits lane's 5c9ed959, in class-v5 from e208dfea on, not in 1c420786's tree: the freeze is the hash object, the worker hosts are the kit's). The fix in flight: the kit's workers (zip 65b47211) on every fleet box, the hive, Windows and Mac packages rebuilt from the kit tree |
## 1. The claim, in one paragraph