igneum-labs
|
4e406ded5c
|
Class v6: the genesis dataset size DECIDED at 4 GiB (the founder's word at 20:00 UK, FROZEN as class_v6_dataset_steps [(0, 4096 MiB)]); change (3) and the served policy sentence as decided; the chip line's a-node-ahead end 3.3x with the floor marked
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 19:21:36 +00:00 |
|
igneum-labs
|
c5d7ee869b
|
Class v6 10.0i: the window core's placed row at about 20:50 (the approximate figures carried); the three-row served line both chip lanes stand behind with the 32-lane floor in
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 19:19:19 +00:00 |
|
igneum-labs
|
1c40e1cb51
|
Class v6 10.0u: the paired 2 GiB read held on PC 1 (the runner held by a fetched job to its cap); the morning's row stands labelled, the read an amendment
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 19:17:53 +00:00 |
|
igneum-labs
|
639904be38
|
Class v6 10.0t: the 32-lane genesis comparator placed (k 0.33 same-node, a floor not a point; the machine 2.1x same-node at the 32-lane floor): the board's same-node bracket 1.5x to 2.1x, the hybrid 1.9x to 2.6x; the served sentence holds it
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 19:15:55 +00:00 |
|
igneum-labs
|
28856b5ff9
|
Class v6 10.0y: F05 corrected on the adversary's placed-core check (the prefix tree never wins on a macro-file core; the literal fold +8.4 percent of the shadow, 2.6 percent of the board; the kill stands); lane 3's package line carried on its branch
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
2026-10-08 19:03:18 +00:00 |
|
igneum-labs
|
7b51b5c348
|
Class v6 10.0x: the B580 rows (11.0 MH/s on class v6 measured, at its random-read ceiling; the watts unread on Level Zero, 10.0 microjoules estimated); the verdict does not turn on them
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
2026-10-08 19:01:03 +00:00 |
|
igneum-labs
|
79a2526cb3
|
Class v6 10.0y: review B's F09(c) (the (c) cell was wrong, every chip fails it, the board six of seven at every ticket), F09(g) (the hybrid operator keeps 85 to 90 percent of the owner's proving surplus; condition (g) holds for the device, not the owner), F05 (the prefix tree 2.6 percent of chip energy; the kill unchanged); lane 3's files at 5cb6b208
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 18:57:34 +00:00 |
|
igneum-labs
|
ca92a8f53b
|
Class v6 10.0x: lane 3's pre-read of ECO-05 at the AMD and Intel cells (a 7600 knee at or under 5.4 adds one sustainable world, 10R at 0.10, and no more; a B580 down to 7 adds none; the envelope unchanged)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 18:45:48 +00:00 |
|
igneum-labs
|
0dfdba82d6
|
Test registry: the research batch rebased on master's 29c87fed (lane 3's eco05 and coexist-fixtures cells own ECO-01, ECO-05 and ECO-08; harness:economics-model keeps the other nine cases), no duplicate cells
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 18:44:45 +00:00 |
|
igneum-labs
|
c4d6a19d13
|
Class v6 10.0x: the AMD-and-Intel energy lane's first rows (the RX 7600 on class v6 15.79 MH/s, 7.2 stock and 5.4 knee modelled, at the ECO-05 threshold; the named cause measured: AMD and Intel cards do 0.22x to 0.47x the 5090's dependent random reads per 32 bits of bus at the same watts); the metered rows by 20:45 and 20:55
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
2026-10-08 18:42:24 +00:00 |
|
igneum-labs
|
f1c4a0f7b6
|
Class v6 10.0i and 10.0h: the k and adversary lanes reconciled on the placed rows (the genesis cores within 6 percent; the bank 0.23x; the machine terms the rest): the line both stand behind, 1.5x to 1.8x the machine and 2.0x to 2.4x the board same-node, 1.8x to 2.1x and 2.45x to 2.8x a node ahead
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 18:40:24 +00:00 |
|
igneum-labs
|
93580ec98c
|
Test registry by the steward's form: three harness cells in the map (economics-model, eco-05, eco-08-package) with coverage per case, the batch eco-d4-20261008-01 (ECO-05 FAIL, the rest RUNNING; manifest 11f6289e), test-record applied, the map page regenerated; the hand edit dropped
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 18:30:57 +00:00 |
|
igneum-labs
|
540b033f40
|
Class v6 10.0i: the first placed gated row (the 32-register base core 7.0 pJ placed, k 0.56 at N3, 2.8x a node ahead and 2.4x node-for-node: the gating and the placement cancel to the digit); the window core placed by 20:15
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 18:29:08 +00:00 |
|
igneum-labs
|
cbb6f95d7b
|
Test registry: the same thirteen rows on the approved copy in its own formatting (a per-case diff, rule 26)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:55:08 +00:00 |
|
igneum-labs
|
fcf55e32ca
|
Test registry: D4's rows re-applied onto the approved copy 534d7020 in its field shape (status, run_status, owner_lane, evidence_path, run_id, updated, evidence_record); lane 3's ECO-05 (RUN: FAIL) and ECO-08 rows taken verbatim
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:53:51 +00:00 |
|
igneum-labs
|
11f6289efd
|
ECO-05: R corrected by a second explorer read (ETC era 6 reward 1.6384 ETC, USD 31.65 M a year; the first page stale); the cube re-run, the verdict unchanged RUN: FAIL; the package and registry carry the corrected R
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:50:15 +00:00 |
|
igneum-labs
|
be15f98fb3
|
ECO-05 RUN: FAIL against the frozen envelope (one sustainable world of sixteen; the AMD, Intel and older NVIDIA classes cost 4x to 6x a Blackwell card per joule on this hash; no specialist passes all three lines; nothing weakened); the ECO-08 package sim/economy/coexist/ (the /tmp import made relative); the registry rows
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:45:14 +00:00 |
|
igneum-labs
|
fc90937b0c
|
Class v6 10.0u: the coordinator's ruling (P03 does not score the genesis size; adversary burden against commodity burden under the entry-tier and 12 GB co-residence constraints): the recommendation to the founder is 4 GiB, 5.5 out on co-residence, 2 GiB the costless alternative; change (3) and the served sentence revised
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:39:20 +00:00 |
|
igneum-labs
|
3fc3493621
|
ECO-05: the scenario register frozen before any result (lane 3 90d1504e: R = ETC's trailing-year miner revenue USD 48.7 M sourced, the P12 worlds verbatim, the envelope verbatim, one BLOCKED item on R); the registry row RUNNING
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:37:43 +00:00 |
|
igneum-labs
|
4c65e3d092
|
Class v6 10.0u: the mapping control (the cost is the size's, not the mapping's; 4 GiB +11.7 percent at the knee, 5.5 GiB +13.4) moves the genesis floor to 2 GiB under P03, grown only by the chain's state; change (3) and the served sentence revised
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:33:46 +00:00 |
|
igneum-labs
|
524ea01bd5
|
Test registry: D4's cases written back (ECO-01 to 04, 06, 07 and INC-01 to 04, 08 RUNNING with their evidence paths; ECO-05 and ECO-08 PLANNED with their clocks); the independent review column open on every row
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:32:24 +00:00 |
|
igneum-labs
|
71c21860e0
|
Reference population: the two Apple rows through lane 3's script (the most efficient honest joule at the GPU rail, the dearest honest owners after the datacentre parts; neither enters); the RX 7600's floor row folded; the floor's honest-side cost per step in one line
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:23:45 +00:00 |
|
igneum-labs
|
931ad47722
|
Class v6 10.0u: the 8 GB tier's own rows (the RX 7600 under 3 percent of rate at 5.5 GiB; the schedule's cost is a tuned-Blackwell cost, not tier-wide); the served sentence narrowed to the card it is true of
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:22:39 +00:00 |
|
igneum-labs
|
5a11af4381
|
Reference population: two team-reported Apple rows (the Mac mini M6 and MacBook Pro M5 Max at 0.73 microjoules per hash on GPU power only, about 3x the locked 5090 at a fraction of the rate; mines, does not prove), their cost cells owed to the script
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:20:59 +00:00 |
|
igneum-labs
|
b200e39190
|
Igneum 2.0: the served reference population and cost-per-accepted-unit table (lane 3 99615191; every cell labelled; the 5070 Ti knee the cheapest owner and entrant; the chip rows beside at the reconciled tickets) in tonight's landing
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:04:10 +00:00 |
|
igneum-labs
|
5656f6eeef
|
Class v6 10.0t: the 32-lane 18-family core's synthesis row (k 0.39 same-node, a sixth under the headline placed; the same-node bracket 1.5x to 2.1x, inside the served sentence)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 17:02:26 +00:00 |
|
igneum-labs
|
1847d41bf0
|
Class v6 10.0u: the dataset schedule scored per step (5.5 GiB +14.3 percent at the Blackwell knee, over the budget, for a chip ticket that flips no verdict; 4 GiB earns its place; later steps state-forced, never scheduled); change (3) revised; the served sentence
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:57:28 +00:00 |
|
igneum-labs
|
0742bc0c8b
|
Class v6 change (1): the best-mix genesis table PASSED lane D's harness (0 hot sets, 0 exhausted, the draw's cost halved against the band; the fold stays the lever for bit R)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:54:13 +00:00 |
|
igneum-labs
|
11f09af39f
|
Class v6 10.3: the W = 8 lock row landed (the lock widens its loss to +11.2 percent of energy; W = 16 +78 percent); both out as regression controls
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:53:13 +00:00 |
|
igneum-labs
|
90523b794f
|
Igneum 2.0 D4: the die at the reconciled USD 1.0 ticket (lane 3's 12c): the verdict moves back, failing (a), (b), (c) and (f) at every point of the band; a sunk USD 10 M fleet 42 percent of a growing chain and 62 of a flat one on landing; the success statement holds at no reconciled ticket
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:26:47 +00:00 |
|
igneum-labs
|
d858442228
|
D4 proving payment: the chain's own block-one read-back (emission 3.168808781 IGN, 2.5350470248 miner and 0.6337617562 pool, 80/20 exact) as the served table's figure; the fee the design, not a payment today
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:25:02 +00:00 |
|
igneum-labs
|
bee747da82
|
Igneum 2.0 D4: the die's ticket reconciled (two machines: zero shadow at USD 0.6, the placed shadow core at 2.94; the adversary's at about USD 1.0 per MH/s, 0.5 to 1.6, set by the power train and the core); a sunk USD 10 M fleet about 10 TH/s; the die's economics are project economics
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:23:50 +00:00 |
|
igneum-labs
|
b16fcc5f02
|
Igneum 2.0 D4: the die's ticket pinned at the adversary lane's USD 2.94 complete machine (the 0.6 row the labelled silicon floor); the die's headline verdict is the hybrid's; the range served until the 18:30 reconciliation
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:19:20 +00:00 |
|
igneum-labs
|
d73e9cfa7f
|
Igneum 2.0 D4: lane 3's re-run at the final energies (12b): the die's verdict moves to the hybrid's at the adversary lane's USD 2.94 ticket and stands at the chip model's 0.6; the 5x ticket gap is the pin that decides it; the board passes six of seven at USD 4.84
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:16:39 +00:00 |
|
igneum-labs
|
1d0d5662ad
|
Igneum 2.0 D4: the final placed energies (the 18-family core: the board 1.5x / 1.8x, the hybrid 1.93x / 2.4x at USD 2.80, the die 2.3x / 3.1x at USD 2.94) and the three chips re-scored: no verdict moves; condition (b) now passes for the hybrid and the die on their corrected tickets, (c) and (f) hold the verdicts; the served sentence 1.5x to 3.1x
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:12:17 +00:00 |
|
igneum-labs
|
bbf9dbfbff
|
Igneum 2.0 D4 complete (lane 3 d0e2018f): the hybrid board scored with the sunk case (holds only in a growing chain at cheap electricity; the floor a real lever on it), the market-structure axis (the board holds under every structure, the die under none but private supply in a growing chain), the sensitivity workbook (171 rows; no threshold served)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:10:33 +00:00 |
|
igneum-labs
|
412a1a22cc
|
Class v6 10.0o: the k lane's synthesised FP32 rows (the FMA unit k 0.65 alone, the blended op 0.15 at the lock: slightly worse forcing work than an integer op); the kill stands
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:06:35 +00:00 |
|
igneum-labs
|
0e35bda6b8
|
Igneum 2.0 D4: the market-structure axis and the sensitivity workbook pinned for 21:00; the success condition to serve, verbatim from the plan
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:05:20 +00:00 |
|
igneum-labs
|
5983d90cfc
|
Igneum 2.0 D4: the placed energies (the complete GDDR7 machine 1.6x node-for-node, 1.9x a node ahead; the hybrid 2.4x / 2.9x; the die 2.4x / 3.3x; per-dollar unchanged) and the three chips scored at them, no verdict changed; the served energy sentence on them; the proving-payment pin in the code behind its constant, the guest's mirror owed
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:03:43 +00:00 |
|
igneum-labs
|
b5ca2bdaa9
|
Igneum 2.0 D4: the coexistence model's second cut and the operator simulation landed (lane 3 a07be1ee: the pass line met for the DRAM board, the SRAM die fails five conditions once sunk; the fixed internal pool is a subsidy, the congestion-priced fee restores a spike in 35 periods); the hybrid board scored on the six conditions (fails the dollar conditions on the first reading)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 16:01:35 +00:00 |
|
igneum-labs
|
919293896e
|
Class v6 10.0h and 10.0i: the sentence served at 18:30 on the tightened bracket (2.5x to 3.0x a node ahead, 2.1x to 2.6x node-for-node; the placed gated row at about 18:30 narrows it)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 15:47:22 +00:00 |
|
igneum-labs
|
ddb0976351
|
Merge the mirror's master into counter-asic-4 before the D4 landing
# Conflicts:
# site/economics.html
|
2026-10-08 15:46:02 +00:00 |
|
igneum-labs
|
9f1c40b683
|
docs: the paid pilot (Igneum 2.0, proving ladder step two): customer profile from the code, three ranked candidate classes with public prices, the pilot shape and pass condition, the missing-pipeline checklist, the outreach brief
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 15:45:41 +00:00 |
|
igneum-labs
|
984aa603d8
|
Igneum 2.0 D4: the tip-share discrepancy resolved (the tip whole to the block, O-5.7 closed at zero; the provers paid by the congestion-priced proving base fee, 90 percent to the block's pool and 10 percent burned; the burn listed separately; the cap and no development tax untouched); spec 5.1 to 5.3, the economics page, the code pinned
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 15:45:00 +00:00 |
|
igneum-labs
|
d846bfe5d8
|
Merge class-v6-connected-docs e6146368 into master (gate: green on e6146368, recorded by tools/ci/pre-push.sh; landed on the box mirror under the exception declared by main: main's ruling, 7 Oct 2026 19:5x UK: the GitHub account is suspended, lanes land on the box mirror's master, the box gate stamp is the verdict; GitHub gets the fast-forward when it answers)
|
2026-10-08 15:43:21 +00:00 |
|
igneum-labs
|
4a3fbecd88
|
Class v6 10.0p: the 8 GB and 12 GB tiers hold at the 5.5 GiB floor (an exact 4060 and 3060, measured); the window free per load on both; proving beside mining time-shares under 16 GB
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 15:40:40 +00:00 |
|
igneum-labs
|
761ec73de0
|
Class v6 10.0o: the mixed FP32 candidate killed (determinism proved on CUDA; the determinism tax is the whole economics, +15 to +26 percent on every card for +7 percent of chip edge; the exponent-byte bias a new instrument reading)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
2026-10-08 15:38:47 +00:00 |
|
igneum-labs
|
5405dd9f20
|
Merge counter-asic-4 03029526 into master (gate: green on 03029526, recorded by tools/ci/pre-push.sh; landed on the build mirror)
|
2026-10-08 15:37:40 +00:00 |
|
igneum-labs
|
9f490a5b8c
|
Merge class-v6-adversary-docs d1d29460 into master (gate: green on d1d29460, recorded by tools/ci/pre-push.sh; landed on the box mirror under the exception declared by main: main's ruling, 7 Oct 2026 19:5x UK: the GitHub account is suspended, lanes land on the box mirror's master, the box gate stamp is the verdict; GitHub gets the fast-forward when it answers)
|
2026-10-08 15:35:33 +00:00 |
|
igneum-labs
|
c67e9dae42
|
Merge release-manifest-8 d5ba3c3a into master (gate: green on d5ba3c3a, recorded by tools/ci/pre-push.sh; landed on the box mirror)
|
2026-10-08 15:32:32 +00:00 |
|