igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity
901 commits 357 branches 1 tag 2 GiB
Commit graph

901 commits

This branch
This branch
All branches
Author SHA1 Message Date
igneum-labs
22f00a91fd Proving v1 plan: fleet night rows 8 to 12 (finality at 93 voters, pool-v0 under load, the 0.3.14 canary, the class v4 rehearsal, the finality pause) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 19:48:55 +00:00
igneum-labs
34ec8fe121 Bench log: the rental cost of hash entry (as cut for the site in d11c98a) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 19:48:09 +00:00
igneum-labs
2e100d7567 Fleet: the class v4 rehearsal tooling (CUDA worker wrapper, generator-4 worker, pack re-export), the standing-fleet rule (supervisor, library, page block, gpu-fleet.md), the 0.3.15 publish script, the Devnet 2 script's ports and bps knobs, the collector's seed knobs 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 19:46:32 +00:00
igneum-labs
72350e992f Block rate on Devnet 2: the plan file (runs A 10 bps and B 1 bps on the 38 wave cards, the readings, the per-tier payout arithmetic) and bps-collect.py; the remote-build tooling from master 2026-10-06 18:01:29 +00:00
igneum-labs
c66abe030b tools/fleet/lib: the one box library (ssh with ServerAlive, scp with retries, install_payload checked on the box, start_node with anchored kills, chain-side readers: height, peers, exec tip, state root, paid segments, rejects and reorgs since a stamp, miner and prover start, destroy with the registry row) and its Devnet 2 test (known-failed and known-finished cases) 2026-10-06 17:58:18 +00:00
igneum-labs
d26a85efcd devnet2-gate: rejects and reorgs counted from the step-1 restart (a joining box's lone-chain reorg is not the release's), bash 3.2 (no mapfile, no associative arrays), the crossing line rewritten 2026-10-06 17:55:13 +00:00
igneum-labs
e3cc2558f5 pgrep-self-match-check: the fleet's scripts anchored (every pattern on the binary path or -x); the live devnet's operational scripts and finished PC playbooks on the owed list by path, named in the README; the check is green over the tree 2026-10-06 17:53:19 +00:00
igneum-labs
6d59105562 fleet: box-matrix.sh kill_server uses pkill -x; the self-match check is green (the previous commit's claim was premature) 2026-10-06 17:47:34 +00:00
igneum-labs
e92956dd56 fleet: box-matrix.sh patterns anchored; pgrep-self-match-check passes over the whole tree 2026-10-06 17:45:08 +00:00
igneum-labs
5e3402b045 fleet: box-node-swap.sh patterns anchored; the self-match check is clean over the tree 2026-10-06 17:43:01 +00:00
igneum-labs
2651ca8cff fleet: the last three bare process patterns anchored; the self-match check passes over the tree 2026-10-06 17:40:30 +00:00
igneum-labs
00675b8e05 fleet scripts: every process pattern anchored (the pgrep-self-match check passes on tools/fleet); the observer's autosync on the owed list 2026-10-06 17:38:23 +00:00
igneum-labs
b97c6edbdd CI: pgrep-self-match-check (a process pattern never matches its own shell; the wave-script class of 6 October), wired into ci.yml, README row; fleet: the 0.3.14 canary (bash 3.2, retries, overridable boxes), Devnet 2 fixes, the wave and pool scripts 2026-10-06 17:35:58 +00:00
igneum-labs
5b95312881 Devnet 2 (the staging chain): box-dn2.sh (igneum-devnet-2, own genesis bits, every activation low, no exec-restart fields), devnet2-override.json, devnet2-gate.sh (install, activation at tip + margin, N minutes, PASS only on zero rejects, reorg depth at most 3, equal exec roots, a paid segment, equal versions), the page's Devnet 2 panel; the wave, pool and snapshot scripts 2026-10-06 16:50:34 +00:00
igneum-labs
d3d460714d proving-v1 plan, the fleet night: the first segment records after the switch accepted and paid (195.8 s, 1.51 IGN); one exec state, the export gap on snapshot-recovered nodes with the 0.3.14 pin roots 2026-10-06 16:43:28 +00:00
igneum-labs
dd0925b98a fleet prover: export from the node's exec restart block, not genesis (the exporter refuses the unexecuted blocks' zero roots on a restarted 0.3.13 node) 2026-10-06 16:10:15 +00:00
igneum-labs
8708f9896b proving-v1 plan, the fleet night: the 0.3.13 swap and replay times per box; DAA 198,000 and the 229-block reorg that reset every exec layer 2026-10-06 15:58:05 +00:00
igneum-labs
d4b187077e fleet: night.py (the prover night loop: launches on the executed tip, re-runs the exec restart on a reset, 15-minute night rows), the tarball sha check in the swap, the prover script's FLOOR line restored 2026-10-06 15:52:59 +00:00
igneum-labs
4bebb6ffb8 fleet: box-node-swap.sh rewritten clean (the edited copy called stamp before its definition and misquoted the sha line; no node was touched) 2026-10-06 15:32:35 +00:00
igneum-labs
48e01c8122 fleet: swap.py (one call per shipper word over the fleet, the hub first), the rig as N provers (per-card loops, distinct sockets), the rig script's device and chain fixes, restage.py 2026-10-06 15:20:00 +00:00
igneum-labs
e1d2462d04 fleet: the node swap's two steps (publish 1 binary with the ten-field file, publish 2 the thirteen-field file), the expected digest checked, igneum_getExecStatus read 2026-10-06 14:59:19 +00:00
igneum-labs
0b61bfa26e fleet: the 0.3.13 node swap for phase-2 boxes (fixed binary over the full-history datadir, the replay timed); the prover launcher prefers the swapped binary 2026-10-06 13:58:53 +00:00
igneum-labs
3618c075fe Prover tiers on real cards: all eleven cards (the 3090 added); bench-log entry complete 2026-10-06 13:28:55 +00:00
igneum-labs
4bee6f04bc bench-log: prover tiers on real cards, the rented fleet (ten cards, the 3090 to follow) 2026-10-06 13:23:37 +00:00
igneum-labs
3154164b97 fleet: the collector keyed by card and VRAM with a raw-rows fallback; the matrix script's final write guarded against a card that reports no power 2026-10-06 13:22:15 +00:00
igneum-labs
c22b4cd504 Prover tiers on real cards: ten cards, the 8 GB beside-the-miner rows (core-only 2^25 fits); fleet: the rig script and retry 2026-10-06 13:20:06 +00:00
igneum-labs
30b2121627 Ember Tune plan: the fleet priors from rented cards (baseline rows; the containers refuse the knobs) 2026-10-06 13:17:48 +00:00
igneum-labs
e380e5075e proving-v1 plan: the fleet night section, the exec finding (a node joining today never executes the chain) as the day's first row, the seed's 30-s drops, the v5 abort gate 2026-10-06 13:15:13 +00:00
igneum-labs
4d04f96ab8 Prover tiers on real cards: eight cards (3060, 3080, 4060 Ti 16 GB, 5090 added), the 8 and 10 GB rows; fleet: the datadir joiner swap, the v5 known-failed runner 2026-10-06 13:07:05 +00:00
igneum-labs
1e884f0ed9 Prover tiers on real cards: the first four cards (4090, A5000, 4070, 5070), the method, the per-tier consequences; fleet: RunPod pods, companions pushed with every stage 2026-10-06 12:56:51 +00:00
igneum-labs
9480489495 GPU fleet: box-kill.sh (anchored kills as a file), log rotation before every stage start, parallel orchestrator probes, the listed-after-rent check 2026-10-06 12:46:14 +00:00
igneum-labs
2e685789b4 GPU fleet: the orchestrator loop, a killable per-second GPU sampler, the STATUS-line rate parser, an atomic locked registry, the p2p hub peer 2026-10-06 12:34:25 +00:00
igneum-labs
bbd2512902 GPU fleet: RunPod client, the phase-1 collector, the phase-2 launcher (node restarted with the proof verifier), parallel status, the toolchain fix 2026-10-06 12:14:57 +00:00
igneum-labs
e78ce96b6f GPU fleet: Vast client, box setup (0.3.12 node + patched SP1 server + cuda host), the phase-1 matrix, the Ember ladder, the Linux segment prover, the fleet page builder 2026-10-06 12:01:23 +00:00
igneum-labs
6442b774d6 morning summary: 12 GB mine-and-prove measured as a core-only prover 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 09:15:27 +00:00
igneum-labs
54b9a78e6c morning summary: Counter ASIC 3.0 closed, the class v4 candidate and its gate decision 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 08:49:04 +00:00
igneum-labs
313b7a2c00 morning summary: the segment rule is a consensus switch; 0.3.12 carries the node 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 08:36:28 +00:00
igneum-labs
5cfc5688f0 morning summary, 6 October 2026 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 08:15:36 +00:00
igneum-labs
6a27d5ab7d release 0.3.11: C35's cause on PC 1 and the ember-tune guard e600e63 in the next-cut list 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 07:03:10 +00:00
igneum-labs
dd0c08515f release 0.3.11: the plan's merge and push section 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 00:14:19 +00:00
igneum-labs
048d50ab89 release 0.3.11: the plan's step 2, the digest sweep and the open items; sections in order 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-06 00:12:23 +00:00
igneum-labs
a9bfa3f3e2 release-0.3.11 plan: the push and both CI runs, step 1 (the hand nodes and the seed at 4d8f8bb6, the manifest with the override carried over, the Mac, PC 2 and its epoch-boundary recovery, the stall and the Mac's resume, the laptop), the next-cut list 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:56:46 +00:00
igneum-labs
9f500d765b igneum-census: build against class v3 (Instr width/win/off, Program class/era_bytes, fetch layout, Scratch/Hot arms) 
The ci run on b2a79ee (37387751432) failed in the census release build only; windows-ci was green.
The census generates the lottery hash's class, so every new field takes its v2 value and the two
experiment ops are unreachable.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:25:52 +00:00
igneum-labs
b2a79ee2f3 Windows payload inputs: node 89dfcb95 (push-inputs.sh, the PC 2 build, the class-aware workers, signed); release-0.3.11 plan: the PC 2 job 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:18:28 +00:00
igneum-labs
081094ec22 release-0.3.11 plan: the three digests on the 0.3.11 binary (4d8f8bb6 with the live file, measured), the two-publish order (C39, C40), the HiveOS package's missing override 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:15:20 +00:00
igneum-labs
a6e0e3f5b2 release-0.3.11 plan: the PC 2 job id, the rollout order as agreed 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:10:03 +00:00
igneum-labs
9b4f5aaed2 release-0.3.11 plan: the relay's PC1 is the 1ccfe586 box (PC 2 restarted twice by my tasks, PC 1 unreachable until the morning), the combined PC 2 job, C38 complete 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:08:52 +00:00
igneum-labs
3c233a8fdf Merge ca2-coord (936f260) into release-0.3.11: the Counter ASIC 2.0 docs commits (N4 rows, gate cells, digests, rules, status; the five cited documents already in the tree) 
# Conflicts:
#	docs/bench-log.md
 2026-10-05 23:06:38 +00:00
igneum-labs
dd74b11a98 prover-floor: the three other PC 2 scripts test their kit file before the first use (C32) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-05 23:06:38 +00:00
igneum-labs
936f26089b Counter ASIC 2.0 status 23:05: C38 closed, the four branches merged into ca2-coord 2026-10-05 23:05:34 +00:00