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

Merge ci-r15-b1 4d2f4925 into master (gate: green on af7124c9, 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)

Browse source
This commit is contained in:
igneum-labs 2026-10-09 10:06:28 +00:00
commit 4fe7e5fac5
7 changed files with 498 additions and 120 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

41
docs/plans/igneum-2.0-test-harness-map.md
View file

@ -483,6 +483,40 @@ Rule: a case maps to a cell only where the cell's tests visibly answer it; cover
- Cases:
- TV-02 Reproduce complete supply accounting: partial: scheduled issuance on a straight chain and its reorg, two independent implementations EQUAL to the integer on devnet-4, mainnet and the TV-01 spec; paid, burned, locked, outstanding, DAG reds, the EVM ledger and duplicate records open; the independent panel has not read it
### model:mhpow-b1-reference
- Command: `tools/mhpow/b1/run-b1.sh <out_dir> all (a build box under its pid file; kat, attack and k-table files byte-identical to tools/mhpow/b1/fixtures/SHA256SUMS)`
- Box class: build box, CPU only (build-6)
- Fixtures: tools/mhpow/b1/fixtures/
- Cases:
- POW-05 Prevent amortised cheap winning attempts: partial: the published primitive's attack fixtures only; the lottery binding and the composed accepted-proof theorem are B2/B4's
- POW-06 Bound verifier work and malformed-input cost: partial: verifier refusals and op counts in CPython; no compiled verifier, no admitted-rate timing
### theory:r15-05-track-b
- Command: `the Track B cryptography read: docs/analysis/mhpow/b1 (the exact CPU reference of the MTP framework over DRSample, its verifier and the six malicious provers) by tools/mhpow/b1/run-b1.sh <out> all on a build box under a pid file, outputs byte-identical to tools/mhpow/b1/fixtures/SHA256SUMS`
- Box class: build box, CPU only
- Fixtures: tools/mhpow/b1/fixtures/
- Cases:
- R15-05 R15-05: Cryptography: partial: the published primitive's reference, attacks and verifier refusals; the lottery binding (B2), the composed theorem (B4) and the panel's reading open
### class:int8
- Command: `the int8-dominated research class: cut behind the research flag by the research lane at int8-50, int8-75, int8-90 and arx-16, arx-32 (igneum-pow export with the class string, on a build box); the hash lane's card rows in the P04 form (tools/1p5x/x1/gpu-pairs.sh, gpu-rows.py); the adversary lane's chip rows (tools/1p5x/x5/sweep.py, tools/1p5x/x8/mac.py); the census instruments (igneum-pow addrsite, x1cache)`
- Box class: build box for the cut and the census; rented pods for the card rows; a build box for the chip model
- Fixtures: F0
- Cases:
- R15-03 R15-03: GPU + memory analysis: partial: the held-out cache curves of each weight against the frozen control; the layer-8-off control is X1's row
- R15-04 R15-04: GPU systems: partial: byte agreement and accepted-work savings of each weight's fold; X2 to X4's rows carry the rest
### model:x7-memory-route
- Command: `the X7 memory-route read: docs/analysis/class-v6/1p5x/x7 (the chip model's s on GDDR7, HBM3 and near-memory by tools/1p5x/x5/sweep.py on a build box under a pid file) against the hash lane's read-only twin row on a rented card at stock (the frozen object and its read-only twin, nvidia-smi beside it)`
- Box class: a build box for the model; a rented card for the twin row
- Fixtures: F0
- Cases:
- R15-02 R15-02: GPU + hardware: partial: the memory route's f and s only; the component power and traffic model and its sensitivity are X0's and X5's rows
## Automated cases with no harness in the matrix (NOT RUN, the reason)
- GOV-02 Approve thresholds before results: the approval is recorded in the registry's approval field; the automated half (thresholds frozen before any run_status) is the gate rule landing by 21:00
@ -491,7 +525,6 @@ Rule: a case maps to a cell only where the cell's tests visibly answer it; cover
- GPU-04 Find the memory-clock operating ladder: the memory-clock ladder has no harness tonight: the project's own rig mines nothing under the the earlier devnet off order
- GPU-06 Measure accepted work under ordinary connectivity: accepted work under ordinary connectivity needs the fault network F4
- GPU-07 Survive sustained thermal and power operation: the sustained thermal and power soak has no harness tonight: the project's own rig mines nothing under the the earlier devnet off order
- POW-05 Prevent amortised cheap winning attempts: amortised cheap winning attempts are the attack lanes' grind and era harnesses (tools/attack/f7-era, f9-grind), not in the release matrix; their rows come from those lanes
- ROT-03 Test miner-voted bring-forward governance: miner-voted bring-forward needs a vote harness on the fault network F4
- ROT-04 Resist seed selection and faster evaluators: seed-selection resistance is the census harness (the class v6 invention lane), not yet in the matrix
- EVM-01 Match the selected EVM semantics: no EVM conformance-vector harness is mapped tonight; the exec suite does not run the reference test vectors
@ -655,10 +688,6 @@ Rule: a case maps to a cell only where the cell's tests visibly answer it; cover
- TV-31 Public-claim and activity review: TV-31 (token value gate, scope CORE): no gate executed; the harness is the owner role's (Qualified counsel); a supplemental layer, never counted as independent evidence; not yet named in the map
- TV-32 Observed leadership, not a roadmap certificate: TV-32 (token value gate, scope LEADERSHIP CLAIM): no gate executed; the harness is the owner role's (Independent acceptance panel); a supplemental layer, never counted as independent evidence; not yet named in the map
- R15-01 R15-01: Release + measurement: R15-01 (1.5x programme package): no package executed; the harness is the lane's (X0); its batch names the case when it lands; not yet named in the map
- R15-02 R15-02: GPU + hardware: R15-02 (1.5x programme package): no package executed; the harness is the lane's (X0, X5); its batch names the case when it lands; not yet named in the map
- R15-03 R15-03: GPU + memory analysis: R15-03 (1.5x programme package): no package executed; the harness is the lane's (X1); its batch names the case when it lands; not yet named in the map
- R15-04 R15-04: GPU systems: R15-04 (1.5x programme package): no package executed; the harness is the lane's (X2, X3, X4); its batch names the case when it lands; not yet named in the map
- R15-05 R15-05: Cryptography: R15-05 (1.5x programme package): no package executed; the harness is the lane's (B1); its batch names the case when it lands; not yet named in the map
- R15-06 R15-06: GPU + cryptography: R15-06 (1.5x programme package): no package executed; the harness is the lane's (B3); its batch names the case when it lands; not yet named in the map
- R15-07 R15-07: Cryptography + hardware: R15-07 (1.5x programme package): no package executed; the harness is the lane's (B2, B4); its batch names the case when it lands; not yet named in the map
- R15-08 R15-08: Independent ASIC team: R15-08 (1.5x programme package): no package executed; the harness is the lane's (X5, B5); its batch names the case when it lands; not yet named in the map
@ -667,4 +696,4 @@ Rule: a case maps to a cell only where the cell's tests visibly answer it; cover
## Count
213 automated cases: 92 mapped to a cell, 179 NOT RUN with a reason.
213 automated cases: 97 mapped to a cell, 174 NOT RUN with a reason.

329
docs/plans/igneum-2.0-test-registry.json
View file

@ -2016,25 +2016,31 @@
"manual_page": 25,
"owner_lane": "pool design seat (a3832b1c3b274b310)",
"run_status": "NOT RUN",
"evidence_path": "docs/analysis/binding-review-2026-10.md (sections 1 to 5a; master ad5a54c5)",
"run_id": "binding-review-2026-10-08-a05",
"updated": "2026-10-08T20:10:24.556Z",
"evidence_path": "docs/analysis/binding-review-2026-10.md (sections 1 to 5a; master ad5a54c5); docs/analysis/mhpow/b1/README.md",
"run_id": "mhpow-b1-20261009-01",
"updated": "2026-10-09T10:06:27.563Z",
"evidence_record": {
"reason": "amortised cheap winning attempts are the attack lanes' grind and era harnesses (tools/attack/f7-era, f9-grind), not in the release matrix; their rows come from those lanes",
"at": "2026-10-08T20:10:24.556Z",
"method": "static",
"requirement_id": "POW-05",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "model:mhpow-b1-reference",
"manifest_sha": "13c3646e",
"run_id": "mhpow-b1-20261009-01",
"evidence": "docs/analysis/mhpow/b1/README.md",
"in_progress": true,
"coverage": "partial: the published primitive's attack fixtures only; the lottery binding and the composed accepted-proof theorem are B2/B4's",
"release_identity": {
"commit": "",
"commit": "13c3646e",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"profile_hashes": ""
}
},
"claim_impact": "none: no public figure moves; Track B's construction is not adopted and nothing activates",
"reviewer": "",
"at": "2026-10-09T10:06:27.563Z",
"note": "POW-05: attack fixtures A1 to A6 (tools/mhpow/b1/fixtures/attacks-n10-k16.json): A1 replay and A2 unopened block REFUSED, A3 control-block steering gain 0 with the red-node fallback accepted at (1-1/t)^k, A4 self-similarity saving refused, A5 Attack 1 accepted at (1-|S|/N)^k, A6 many accepted roots per labelling (the lottery binding BLOCKED to B2/B4, rows O-1 and O-8). POW-06: the verifier's named checks refuse malformed containers, wrong positions, skipped parents and wrong tags (test_b1.py); proof bytes and verifier oracle calls per N are exact formulas with measured rows; cold verification at rate BLOCKED to B6 (O-11). | B1 (R15-05), the cryptography lane of the 1.5x programme's Track B (9 October 2026): the exact CPU reference of Blocki and Smearsoll's MTP framework (eprint 2025/1456 as archived, sha256 13c3646e..., Sec 2.2, 3.2, App. C) over DRSample (ABH17 Alg. 1, eprint 2017/443 as archived, sha256 f8e08d36...), its complete verifier (named checks in the paper's order, each with a known-failed test, 17 of 17 OK on build-6), five deterministic known-answer fixtures (n 4 to 20), honest-prover rows n 12 to 24 (CPython, wall time only, no power) and six malicious provers (Dinur-Nadler eprint 2017/497 Sec 4.1, 4.3/5, 8; the paper's Attack 1 and LuckyQuery regrind). Dinur-Nadler's steering and unopened-block attacks are refused; red-node cheats are accepted with the (1-|S|/N)^k probability the paper prices; one labelling yields many accepted roots (F-4). The proven k with ABH17's constant exceeds N to about n 25 (F-3). NOT RUN until the panel reads it; this lane writes no PASS."
},
"status_note": "REVIEWED: fixed by construction (binding), not applicable (final digest); native tests ok on build-2; no change for 2.0",
"approvals": {
@ -2065,8 +2071,32 @@
"run_id": "binding-review-2026-10-08-a05",
"evidence": "docs/analysis/binding-review-2026-10.md (sections 1 to 5a; master ad5a54c5)",
"in_progress": false
},
"model:mhpow-b1-reference": {
"requirement_id": "POW-05",
"decision": "NOT RUN",
"method": "model",
"cell": "model:mhpow-b1-reference",
"manifest_sha": "13c3646e",
"run_id": "mhpow-b1-20261009-01",
"evidence": "docs/analysis/mhpow/b1/README.md",
"in_progress": true,
"coverage": "partial: the published primitive's attack fixtures only; the lottery binding and the composed accepted-proof theorem are B2/B4's",
"release_identity": {
"commit": "13c3646e",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"profile_hashes": ""
},
"claim_impact": "none: no public figure moves; Track B's construction is not adopted and nothing activates",
"reviewer": "",
"at": "2026-10-09T10:06:27.563Z",
"note": "POW-05: attack fixtures A1 to A6 (tools/mhpow/b1/fixtures/attacks-n10-k16.json): A1 replay and A2 unopened block REFUSED, A3 control-block steering gain 0 with the red-node fallback accepted at (1-1/t)^k, A4 self-similarity saving refused, A5 Attack 1 accepted at (1-|S|/N)^k, A6 many accepted roots per labelling (the lottery binding BLOCKED to B2/B4, rows O-1 and O-8). POW-06: the verifier's named checks refuse malformed containers, wrong positions, skipped parents and wrong tags (test_b1.py); proof bytes and verifier oracle calls per N are exact formulas with measured rows; cold verification at rate BLOCKED to B6 (O-11). | B1 (R15-05), the cryptography lane of the 1.5x programme's Track B (9 October 2026): the exact CPU reference of Blocki and Smearsoll's MTP framework (eprint 2025/1456 as archived, sha256 13c3646e..., Sec 2.2, 3.2, App. C) over DRSample (ABH17 Alg. 1, eprint 2017/443 as archived, sha256 f8e08d36...), its complete verifier (named checks in the paper's order, each with a known-failed test, 17 of 17 OK on build-6), five deterministic known-answer fixtures (n 4 to 20), honest-prover rows n 12 to 24 (CPython, wall time only, no power) and six malicious provers (Dinur-Nadler eprint 2017/497 Sec 4.1, 4.3/5, 8; the paper's Attack 1 and LuckyQuery regrind). Dinur-Nadler's steering and unopened-block attacks are refused; red-node cheats are accepted with the (1-|S|/N)^k probability the paper prices; one labelling yields many accepted roots (F-4). The proven k with ABH17's constant exceeds N to about n 25 (F-3). NOT RUN until the panel reads it; this lane writes no PASS."
}
}
},
"in_progress_since": "2026-10-09T10:06:27.563Z"
},
{
"id": "POW-06",
@ -2095,29 +2125,31 @@
"manual_page": 25,
"owner_lane": "hash lane (a690540514aa453d7)",
"run_status": "NOT RUN",
"evidence_path": "build-1:/srv/artefacts/tas/p01-partb-20261008-01/summary.jsonl",
"run_id": "p01-partb-20261008-01",
"updated": "2026-10-08T18:28:32.463Z",
"evidence_path": "build-1:/srv/artefacts/tas/p01-partb-20261008-01/summary.jsonl; docs/analysis/mhpow/b1/README.md",
"run_id": "mhpow-b1-20261009-01",
"updated": "2026-10-09T10:06:27.563Z",
"evidence_record": {
"cell": "harness:parser-malformed",
"manifest_sha": "417c4a57",
"coverage": {
"POW-06": "partial: three parsers of external bytes (the state stream, the igneum_ RPC params, the pool job line), ten thousand seeded malformed cases each, zero panics, every case under 500 ms (the slowest 152 microseconds); 'accepted' counts mutations that left a valid input, not a parser fault; the verifier work bound (the other half of the accept text) is the proving lane's measurement"
},
"at": "2026-10-08T18:28:32.463Z",
"method": "static",
"requirement_id": "POW-06",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "model:mhpow-b1-reference",
"manifest_sha": "13c3646e",
"run_id": "mhpow-b1-20261009-01",
"evidence": "docs/analysis/mhpow/b1/README.md",
"in_progress": true,
"coverage": "partial: verifier refusals and op counts in CPython; no compiled verifier, no admitted-rate timing",
"release_identity": {
"commit": "417c4a57",
"commit": "13c3646e",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"profile_hashes": ""
}
},
"claim_impact": "none: no public figure moves; Track B's construction is not adopted and nothing activates",
"reviewer": "",
"at": "2026-10-09T10:06:27.563Z",
"note": "POW-05: attack fixtures A1 to A6 (tools/mhpow/b1/fixtures/attacks-n10-k16.json): A1 replay and A2 unopened block REFUSED, A3 control-block steering gain 0 with the red-node fallback accepted at (1-1/t)^k, A4 self-similarity saving refused, A5 Attack 1 accepted at (1-|S|/N)^k, A6 many accepted roots per labelling (the lottery binding BLOCKED to B2/B4, rows O-1 and O-8). POW-06: the verifier's named checks refuse malformed containers, wrong positions, skipped parents and wrong tags (test_b1.py); proof bytes and verifier oracle calls per N are exact formulas with measured rows; cold verification at rate BLOCKED to B6 (O-11). | B1 (R15-05), the cryptography lane of the 1.5x programme's Track B (9 October 2026): the exact CPU reference of Blocki and Smearsoll's MTP framework (eprint 2025/1456 as archived, sha256 13c3646e..., Sec 2.2, 3.2, App. C) over DRSample (ABH17 Alg. 1, eprint 2017/443 as archived, sha256 f8e08d36...), its complete verifier (named checks in the paper's order, each with a known-failed test, 17 of 17 OK on build-6), five deterministic known-answer fixtures (n 4 to 20), honest-prover rows n 12 to 24 (CPython, wall time only, no power) and six malicious provers (Dinur-Nadler eprint 2017/497 Sec 4.1, 4.3/5, 8; the paper's Attack 1 and LuckyQuery regrind). Dinur-Nadler's steering and unopened-block attacks are refused; red-node cheats are accepted with the (1-|S|/N)^k probability the paper prices; one labelling yields many accepted roots (F-4). The proven k with ABH17's constant exceeds N to about n 25 (F-3). NOT RUN until the panel reads it; this lane writes no PASS."
},
"in_progress_since": "2026-10-08T18:28:32.463Z",
"approvals": {
@ -2150,6 +2182,29 @@
"run_id": "p01-partb-20261008-01",
"evidence": "build-1:/srv/artefacts/tas/p01-partb-20261008-01/summary.jsonl",
"in_progress": true
},
"model:mhpow-b1-reference": {
"requirement_id": "POW-06",
"decision": "NOT RUN",
"method": "model",
"cell": "model:mhpow-b1-reference",
"manifest_sha": "13c3646e",
"run_id": "mhpow-b1-20261009-01",
"evidence": "docs/analysis/mhpow/b1/README.md",
"in_progress": true,
"coverage": "partial: verifier refusals and op counts in CPython; no compiled verifier, no admitted-rate timing",
"release_identity": {
"commit": "13c3646e",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"profile_hashes": ""
},
"claim_impact": "none: no public figure moves; Track B's construction is not adopted and nothing activates",
"reviewer": "",
"at": "2026-10-09T10:06:27.563Z",
"note": "POW-05: attack fixtures A1 to A6 (tools/mhpow/b1/fixtures/attacks-n10-k16.json): A1 replay and A2 unopened block REFUSED, A3 control-block steering gain 0 with the red-node fallback accepted at (1-1/t)^k, A4 self-similarity saving refused, A5 Attack 1 accepted at (1-|S|/N)^k, A6 many accepted roots per labelling (the lottery binding BLOCKED to B2/B4, rows O-1 and O-8). POW-06: the verifier's named checks refuse malformed containers, wrong positions, skipped parents and wrong tags (test_b1.py); proof bytes and verifier oracle calls per N are exact formulas with measured rows; cold verification at rate BLOCKED to B6 (O-11). | B1 (R15-05), the cryptography lane of the 1.5x programme's Track B (9 October 2026): the exact CPU reference of Blocki and Smearsoll's MTP framework (eprint 2025/1456 as archived, sha256 13c3646e..., Sec 2.2, 3.2, App. C) over DRSample (ABH17 Alg. 1, eprint 2017/443 as archived, sha256 f8e08d36...), its complete verifier (named checks in the paper's order, each with a known-failed test, 17 of 17 OK on build-6), five deterministic known-answer fixtures (n 4 to 20), honest-prover rows n 12 to 24 (CPython, wall time only, no power) and six malicious provers (Dinur-Nadler eprint 2017/497 Sec 4.1, 4.3/5, 8; the paper's Attack 1 and LuckyQuery regrind). Dinur-Nadler's steering and unopened-block attacks are refused; red-node cheats are accepted with the (1-|S|/N)^k probability the paper prices; one labelling yields many accepted roots (F-4). The proven k with ABH17's constant exceeds N to about n 25 (F-3). NOT RUN until the panel reads it; this lane writes no PASS."
}
}
},
@ -20321,42 +20376,53 @@
"manual_page": null,
"run_status": "NOT RUN",
"plan_status": "NOT RUN",
"updated": "2026-10-09T09:19:24.923Z",
"updated": "2026-10-09T10:06:27.649Z",
"evidence_record": {
"reason": "R15-02 (1.5x programme package): no package executed; the harness is the lane's (X0, X5); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"requirement_id": "R15-02",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "model:x7-memory-route",
"manifest_sha": "6ba3f01b",
"run_id": "r15-02-x7-20261009-01",
"evidence": "docs/analysis/class-v6/1p5x/x7/README.md; tools/ci/batches/adversary-20261009-x7.json",
"in_progress": true,
"coverage": "partial: the memory route's f and s only; the component power and traffic model and its sensitivity are X0's and X5's rows",
"release_identity": {
"commit": "",
"commit": "master 6ba3f01b (the X7 README; the verdict commit ca2fb6b2 and X8's rows e64eb265 cited when the landing hand lands them)",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
}
},
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.649Z",
"note": "BOUND-side negative control (X7): f = 0.854 on the 4090 at stock, the frozen object at 8,077 nJ per hash against its read-only twin at 6,899 at the same rate (the hash lane's row, fk-x7-4090, 10:44 UK); s = 13 to 39 on the chip's GDDR7, HBM3 and near-memory (the adversary lane's X7 rows): no read-dominated design reaches 1.5x or 1.25x at any f, because the card's random-read path is its least efficient part against a chip. The chip figures are modelled (BOUND); the card figure is a WITNESS on a named rented card. X8's chip side (every int8 mix raising the ratio against a systolic array; the dp4a-only cells stopping at 1.54x) is the tensor route's first witness row set, pending the hash lane's measured card rows; recorded here when e64eb265 lands. R15-02 (no device or whole-machine boundary mismatch) stays NOT RUN until read | main's order through the coordinator (11:0x UK, 9 October 2026): X7's memory-route verdict recorded as a BOUND-side negative control under R15-02, NOT RUN; X8's tensor-route chip rows join when they land and the hash lane's measured card rows arrive"
},
"evidence_records": {
"record": {
"reason": "R15-02 (1.5x programme package): no package executed; the harness is the lane's (X0, X5); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"model:x7-memory-route": {
"requirement_id": "R15-02",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "model:x7-memory-route",
"manifest_sha": "6ba3f01b",
"run_id": "r15-02-x7-20261009-01",
"evidence": "docs/analysis/class-v6/1p5x/x7/README.md; tools/ci/batches/adversary-20261009-x7.json",
"in_progress": true,
"coverage": "partial: the memory route's f and s only; the component power and traffic model and its sensitivity are X0's and X5's rows",
"release_identity": {
"commit": "",
"commit": "master 6ba3f01b (the X7 README; the verdict commit ca2fb6b2 and X8's rows e64eb265 cited when the landing hand lands them)",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"in_progress": false
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.649Z",
"note": "BOUND-side negative control (X7): f = 0.854 on the 4090 at stock, the frozen object at 8,077 nJ per hash against its read-only twin at 6,899 at the same rate (the hash lane's row, fk-x7-4090, 10:44 UK); s = 13 to 39 on the chip's GDDR7, HBM3 and near-memory (the adversary lane's X7 rows): no read-dominated design reaches 1.5x or 1.25x at any f, because the card's random-read path is its least efficient part against a chip. The chip figures are modelled (BOUND); the card figure is a WITNESS on a named rented card. X8's chip side (every int8 mix raising the ratio against a systolic array; the dp4a-only cells stopping at 1.54x) is the tensor route's first witness row set, pending the hash lane's measured card rows; recorded here when e64eb265 lands. R15-02 (no device or whole-machine boundary mismatch) stays NOT RUN until read | main's order through the coordinator (11:0x UK, 9 October 2026): X7's memory-route verdict recorded as a BOUND-side negative control under R15-02, NOT RUN; X8's tensor-route chip rows join when they land and the hash lane's measured card rows arrive"
}
},
"approvals": {
@ -20364,7 +20430,10 @@
"implementation_complete": null,
"evidence_reproduced": null,
"claim_authorised": null
}
},
"run_id": "r15-02-x7-20261009-01",
"evidence_path": "docs/analysis/class-v6/1p5x/x7/README.md; tools/ci/batches/adversary-20261009-x7.json",
"in_progress_since": "2026-10-09T10:06:27.649Z"
},
{
"id": "R15-03",
@ -20397,42 +20466,53 @@
"manual_page": null,
"run_status": "NOT RUN",
"plan_status": "NOT RUN",
"updated": "2026-10-09T09:19:24.923Z",
"updated": "2026-10-09T10:06:27.734Z",
"evidence_record": {
"reason": "R15-03 (1.5x programme package): no package executed; the harness is the lane's (X1); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"requirement_id": "R15-03",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "class:int8",
"manifest_sha": "d799c086",
"run_id": "r15-int8-20261009-01",
"evidence": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"in_progress": true,
"coverage": "partial: the held-out cache curves of each weight against the frozen control; the layer-8-off control is X1's row",
"release_identity": {
"commit": "",
"commit": "master d799c086; the class is cut by the research lane behind a research flag, not in any release",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
}
},
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.734Z",
"note": "opened, nothing measured yet: the int8-dominated class behind a research flag at three weights (int8-50, int8-75, int8-90) and the ARX-only points (arx-16, arx-32); the research lane cuts the class, the hash lane measures the card rows, the adversary lane the chip rows, the census instruments read it; R15-03 (layer-8-off control and held-out cache curves) and R15-04 (compaction, exact-fold alternatives, tuning) take their rows from those batches | main's order through the coordinator (11:0x UK, 9 October 2026): the cells for the int8-dominated class opened, every row NOT RUN until read; the lanes land their own batches on these cells with map_cell_requested"
},
"evidence_records": {
"record": {
"reason": "R15-03 (1.5x programme package): no package executed; the harness is the lane's (X1); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"class:int8": {
"requirement_id": "R15-03",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "class:int8",
"manifest_sha": "d799c086",
"run_id": "r15-int8-20261009-01",
"evidence": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"in_progress": true,
"coverage": "partial: the held-out cache curves of each weight against the frozen control; the layer-8-off control is X1's row",
"release_identity": {
"commit": "",
"commit": "master d799c086; the class is cut by the research lane behind a research flag, not in any release",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"in_progress": false
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.734Z",
"note": "opened, nothing measured yet: the int8-dominated class behind a research flag at three weights (int8-50, int8-75, int8-90) and the ARX-only points (arx-16, arx-32); the research lane cuts the class, the hash lane measures the card rows, the adversary lane the chip rows, the census instruments read it; R15-03 (layer-8-off control and held-out cache curves) and R15-04 (compaction, exact-fold alternatives, tuning) take their rows from those batches | main's order through the coordinator (11:0x UK, 9 October 2026): the cells for the int8-dominated class opened, every row NOT RUN until read; the lanes land their own batches on these cells with map_cell_requested"
}
},
"approvals": {
@ -20440,7 +20520,10 @@
"implementation_complete": null,
"evidence_reproduced": null,
"claim_authorised": null
}
},
"run_id": "r15-int8-20261009-01",
"evidence_path": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"in_progress_since": "2026-10-09T10:06:27.734Z"
},
{
"id": "R15-04",
@ -20483,42 +20566,53 @@
"manual_page": null,
"run_status": "NOT RUN",
"plan_status": "NOT RUN",
"updated": "2026-10-09T09:19:24.923Z",
"updated": "2026-10-09T10:06:27.734Z",
"evidence_record": {
"reason": "R15-04 (1.5x programme package): no package executed; the harness is the lane's (X2, X3, X4); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"requirement_id": "R15-04",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "class:int8",
"manifest_sha": "d799c086",
"run_id": "r15-int8-20261009-01",
"evidence": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"in_progress": true,
"coverage": "partial: byte agreement and accepted-work savings of each weight's fold; X2 to X4's rows carry the rest",
"release_identity": {
"commit": "",
"commit": "master d799c086; the class is cut by the research lane behind a research flag, not in any release",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
}
},
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.734Z",
"note": "opened, nothing measured yet: the int8-dominated class behind a research flag at three weights (int8-50, int8-75, int8-90) and the ARX-only points (arx-16, arx-32); the research lane cuts the class, the hash lane measures the card rows, the adversary lane the chip rows, the census instruments read it; R15-03 (layer-8-off control and held-out cache curves) and R15-04 (compaction, exact-fold alternatives, tuning) take their rows from those batches | main's order through the coordinator (11:0x UK, 9 October 2026): the cells for the int8-dominated class opened, every row NOT RUN until read; the lanes land their own batches on these cells with map_cell_requested"
},
"evidence_records": {
"record": {
"reason": "R15-04 (1.5x programme package): no package executed; the harness is the lane's (X2, X3, X4); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"class:int8": {
"requirement_id": "R15-04",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "class:int8",
"manifest_sha": "d799c086",
"run_id": "r15-int8-20261009-01",
"evidence": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"in_progress": true,
"coverage": "partial: byte agreement and accepted-work savings of each weight's fold; X2 to X4's rows carry the rest",
"release_identity": {
"commit": "",
"commit": "master d799c086; the class is cut by the research lane behind a research flag, not in any release",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"in_progress": false
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.734Z",
"note": "opened, nothing measured yet: the int8-dominated class behind a research flag at three weights (int8-50, int8-75, int8-90) and the ARX-only points (arx-16, arx-32); the research lane cuts the class, the hash lane measures the card rows, the adversary lane the chip rows, the census instruments read it; R15-03 (layer-8-off control and held-out cache curves) and R15-04 (compaction, exact-fold alternatives, tuning) take their rows from those batches | main's order through the coordinator (11:0x UK, 9 October 2026): the cells for the int8-dominated class opened, every row NOT RUN until read; the lanes land their own batches on these cells with map_cell_requested"
}
},
"approvals": {
@ -20526,7 +20620,10 @@
"implementation_complete": null,
"evidence_reproduced": null,
"claim_authorised": null
}
},
"run_id": "r15-int8-20261009-01",
"evidence_path": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"in_progress_since": "2026-10-09T10:06:27.734Z"
},
{
"id": "R15-05",
@ -20559,42 +20656,53 @@
"manual_page": null,
"run_status": "NOT RUN",
"plan_status": "NOT RUN",
"updated": "2026-10-09T09:19:24.923Z",
"updated": "2026-10-09T10:06:27.821Z",
"evidence_record": {
"reason": "R15-05 (1.5x programme package): no package executed; the harness is the lane's (B1); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"requirement_id": "R15-05",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "theory:r15-05-track-b",
"manifest_sha": "0f370527",
"run_id": "r15-track-b-20261009-02",
"evidence": "docs/analysis/mhpow/b1/README.md; tools/mhpow/b1/fixtures/attacks-n10-k16.json; tools/mhpow/b1/fixtures/SHA256SUMS",
"in_progress": true,
"coverage": "partial: the published primitive's reference, attacks and verifier refusals; the lottery binding (B2), the composed theorem (B4) and the panel's reading open",
"release_identity": {
"commit": "",
"commit": "master 0f370527 (B1 e9f94df6 landed)",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
}
},
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.821Z",
"note": "BOUND (B1, F-3): with ABH17's proven constant the paper's challenge count k exceeds N at lambda 256 and N 2^20 (to about n 25): no practical parameter in the proven regime. FINDING (B1, F-4): one labelling yields 193 to 200 accepted roots in 200 tries at 26 oracle calls per try, so a lottery over tau is not bound to the labelling cost (fixture A6, attacks-n10-k16.json, run on build-6). The package R15-05 (no improvised sampled-memory soundness) stays NOT RUN until the independent panel reads the reference and its verifier (17 of 17 named checks with known-failed tests) | the coordinator's order (10:4x UK, 9 October 2026): Track B's reproduction findings as BOUND rows, NOT RUN, no pass or fail claimed for the package"
},
"evidence_records": {
"record": {
"reason": "R15-05 (1.5x programme package): no package executed; the harness is the lane's (B1); its batch names the case when it lands; not yet named in the map",
"at": "2026-10-09T09:19:24.923Z",
"method": "static",
"theory:r15-05-track-b": {
"requirement_id": "R15-05",
"decision": "NOT RUN",
"reviewer": "",
"claim_impact": "",
"method": "model",
"cell": "theory:r15-05-track-b",
"manifest_sha": "0f370527",
"run_id": "r15-track-b-20261009-02",
"evidence": "docs/analysis/mhpow/b1/README.md; tools/mhpow/b1/fixtures/attacks-n10-k16.json; tools/mhpow/b1/fixtures/SHA256SUMS",
"in_progress": true,
"coverage": "partial: the published primitive's reference, attacks and verifier refusals; the lottery binding (B2), the composed theorem (B4) and the panel's reading open",
"release_identity": {
"commit": "",
"commit": "master 0f370527 (B1 e9f94df6 landed)",
"lockfile": "",
"binary": "",
"network_object": "",
"activation": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"in_progress": false
"claim_impact": "none",
"reviewer": "",
"at": "2026-10-09T10:06:27.821Z",
"note": "BOUND (B1, F-3): with ABH17's proven constant the paper's challenge count k exceeds N at lambda 256 and N 2^20 (to about n 25): no practical parameter in the proven regime. FINDING (B1, F-4): one labelling yields 193 to 200 accepted roots in 200 tries at 26 oracle calls per try, so a lottery over tau is not bound to the labelling cost (fixture A6, attacks-n10-k16.json, run on build-6). The package R15-05 (no improvised sampled-memory soundness) stays NOT RUN until the independent panel reads the reference and its verifier (17 of 17 named checks with known-failed tests) | the coordinator's order (10:4x UK, 9 October 2026): Track B's reproduction findings as BOUND rows, NOT RUN, no pass or fail claimed for the package"
}
},
"approvals": {
@ -20602,7 +20710,10 @@
"implementation_complete": null,
"evidence_reproduced": null,
"claim_authorised": null
}
},
"run_id": "r15-track-b-20261009-02",
"evidence_path": "docs/analysis/mhpow/b1/README.md; tools/mhpow/b1/fixtures/attacks-n10-k16.json; tools/mhpow/b1/fixtures/SHA256SUMS",
"in_progress_since": "2026-10-09T10:06:27.821Z"
},
{
"id": "R15-06",

42
tools/ci/batches/mhpow-b1-20261009-01.json Normal file
View file

@ -0,0 +1,42 @@
{
"run_id": "mhpow-b1-20261009-01",
"manifest_sha": "13c3646e",
"evidence_dir": "docs/analysis/mhpow/b1",
"method": "model",
"boxes": [
"build-6"
],
"note": "B1 (R15-05), the cryptography lane of the 1.5x programme's Track B (9 October 2026): the exact CPU reference of Blocki and Smearsoll's MTP framework (eprint 2025/1456 as archived, sha256 13c3646e..., Sec 2.2, 3.2, App. C) over DRSample (ABH17 Alg. 1, eprint 2017/443 as archived, sha256 f8e08d36...), its complete verifier (named checks in the paper's order, each with a known-failed test, 17 of 17 OK on build-6), five deterministic known-answer fixtures (n 4 to 20), honest-prover rows n 12 to 24 (CPython, wall time only, no power) and six malicious provers (Dinur-Nadler eprint 2017/497 Sec 4.1, 4.3/5, 8; the paper's Attack 1 and LuckyQuery regrind). Dinur-Nadler's steering and unopened-block attacks are refused; red-node cheats are accepted with the (1-|S|/N)^k probability the paper prices; one labelling yields many accepted roots (F-4). The proven k with ABH17's constant exceeds N to about n 25 (F-3). NOT RUN until the panel reads it; this lane writes no PASS.",
"cells": [
{
"cell": "model:mhpow-b1-reference",
"cases": [
"POW-05",
"POW-06"
],
"status": "NOT RUN",
"method": "model",
"evidence": "docs/analysis/mhpow/b1/README.md",
"note": "POW-05: attack fixtures A1 to A6 (tools/mhpow/b1/fixtures/attacks-n10-k16.json): A1 replay and A2 unopened block REFUSED, A3 control-block steering gain 0 with the red-node fallback accepted at (1-1/t)^k, A4 self-similarity saving refused, A5 Attack 1 accepted at (1-|S|/N)^k, A6 many accepted roots per labelling (the lottery binding BLOCKED to B2/B4, rows O-1 and O-8). POW-06: the verifier's named checks refuse malformed containers, wrong positions, skipped parents and wrong tags (test_b1.py); proof bytes and verifier oracle calls per N are exact formulas with measured rows; cold verification at rate BLOCKED to B6 (O-11).",
"claim_impact": "none: no public figure moves; Track B's construction is not adopted and nothing activates",
"in_progress": true
}
],
"map_cell_requested": {
"model:mhpow-b1-reference": {
"command": "tools/mhpow/b1/run-b1.sh <out_dir> all (a build box under its pid file; kat, attack and k-table files byte-identical to tools/mhpow/b1/fixtures/SHA256SUMS)",
"box_class": "build box, CPU only (build-6)",
"fixtures": [
"tools/mhpow/b1/fixtures/"
],
"cases": [
"POW-05",
"POW-06"
],
"coverage": {
"POW-05": "partial: the published primitive's attack fixtures only; the lottery binding and the composed accepted-proof theorem are B2/B4's",
"POW-06": "partial: verifier refusals and op counts in CPython; no compiled verifier, no admitted-rate timing"
}
}
}
}

48
tools/ci/batches/r15-02-x7-20261009-01.json Normal file
View file

@ -0,0 +1,48 @@
{
"run_id": "r15-02-x7-20261009-01",
"manifest_sha": "6ba3f01b",
"method": "model",
"evidence_dir": "docs/analysis/class-v6/1p5x/x7",
"boxes": [
"build-9"
],
"release_identity": {
"commit": "master 6ba3f01b (the X7 README; the verdict commit ca2fb6b2 and X8's rows e64eb265 cited when the landing hand lands them)",
"lockfile": "",
"binary": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"claim_impact": "none: a negative control on the memory route; no served figure moves",
"note": "main's order through the coordinator (11:0x UK, 9 October 2026): X7's memory-route verdict recorded as a BOUND-side negative control under R15-02, NOT RUN; X8's tensor-route chip rows join when they land and the hash lane's measured card rows arrive",
"cells": [
{
"cell": "model:x7-memory-route",
"cases": [
"R15-02"
],
"status": "NOT RUN",
"in_progress": true,
"method": "model",
"evidence": "docs/analysis/class-v6/1p5x/x7/README.md; tools/ci/batches/adversary-20261009-x7.json",
"note": "BOUND-side negative control (X7): f = 0.854 on the 4090 at stock, the frozen object at 8,077 nJ per hash against its read-only twin at 6,899 at the same rate (the hash lane's row, fk-x7-4090, 10:44 UK); s = 13 to 39 on the chip's GDDR7, HBM3 and near-memory (the adversary lane's X7 rows): no read-dominated design reaches 1.5x or 1.25x at any f, because the card's random-read path is its least efficient part against a chip. The chip figures are modelled (BOUND); the card figure is a WITNESS on a named rented card. X8's chip side (every int8 mix raising the ratio against a systolic array; the dp4a-only cells stopping at 1.54x) is the tensor route's first witness row set, pending the hash lane's measured card rows; recorded here when e64eb265 lands. R15-02 (no device or whole-machine boundary mismatch) stays NOT RUN until read",
"claim_impact": "none"
}
],
"map_cell_requested": {
"model:x7-memory-route": {
"command": "the X7 memory-route read: docs/analysis/class-v6/1p5x/x7 (the chip model's s on GDDR7, HBM3 and near-memory by tools/1p5x/x5/sweep.py on a build box under a pid file) against the hash lane's read-only twin row on a rented card at stock (the frozen object and its read-only twin, nvidia-smi beside it)",
"box_class": "a build box for the model; a rented card for the twin row",
"fixtures": [
"F0"
],
"cases": [
"R15-02"
],
"coverage": {
"R15-02": "partial: the memory route's f and s only; the component power and traffic model and its sensitivity are X0's and X5's rows"
}
}
}
}

49
tools/ci/batches/r15-int8-20261009-01.json Normal file
View file

@ -0,0 +1,49 @@
{
"run_id": "r15-int8-20261009-01",
"manifest_sha": "d799c086",
"method": "model",
"evidence_dir": "docs/plans/igneum-2.0-master/1p5x/research-plan",
"boxes": [],
"release_identity": {
"commit": "master d799c086; the class is cut by the research lane behind a research flag, not in any release",
"lockfile": "",
"binary": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"claim_impact": "none: a research class behind a flag; no served figure moves and nothing activates",
"note": "main's order through the coordinator (11:0x UK, 9 October 2026): the cells for the int8-dominated class opened, every row NOT RUN until read; the lanes land their own batches on these cells with map_cell_requested",
"cells": [
{
"cell": "class:int8",
"cases": [
"R15-03",
"R15-04"
],
"status": "NOT RUN",
"in_progress": true,
"method": "model",
"evidence": "docs/plans/igneum-2.0-master/1p5x/research-plan/packages.json; docs/analysis/class-v6/1p5x/x1/README.md",
"note": "opened, nothing measured yet: the int8-dominated class behind a research flag at three weights (int8-50, int8-75, int8-90) and the ARX-only points (arx-16, arx-32); the research lane cuts the class, the hash lane measures the card rows, the adversary lane the chip rows, the census instruments read it; R15-03 (layer-8-off control and held-out cache curves) and R15-04 (compaction, exact-fold alternatives, tuning) take their rows from those batches",
"claim_impact": "none"
}
],
"map_cell_requested": {
"class:int8": {
"command": "the int8-dominated research class: cut behind the research flag by the research lane at int8-50, int8-75, int8-90 and arx-16, arx-32 (igneum-pow export with the class string, on a build box); the hash lane's card rows in the P04 form (tools/1p5x/x1/gpu-pairs.sh, gpu-rows.py); the adversary lane's chip rows (tools/1p5x/x5/sweep.py, tools/1p5x/x8/mac.py); the census instruments (igneum-pow addrsite, x1cache)",
"box_class": "build box for the cut and the census; rented pods for the card rows; a build box for the chip model",
"fixtures": [
"F0"
],
"cases": [
"R15-03",
"R15-04"
],
"coverage": {
"R15-03": "partial: the held-out cache curves of each weight against the frozen control; the layer-8-off control is X1's row",
"R15-04": "partial: byte agreement and accepted-work savings of each weight's fold; X2 to X4's rows carry the rest"
}
}
}
}

48
tools/ci/batches/r15-track-b-20261009-02.json Normal file
View file

@ -0,0 +1,48 @@
{
"run_id": "r15-track-b-20261009-02",
"manifest_sha": "0f370527",
"method": "model",
"evidence_dir": "docs/analysis/mhpow/b1",
"boxes": [
"build-6"
],
"release_identity": {
"commit": "master 0f370527 (B1 e9f94df6 landed)",
"lockfile": "",
"binary": "",
"network_object": "none: documents and models only, no chain state read or written",
"activation": "none",
"profile_hashes": ""
},
"claim_impact": "none: no served figure moves; Track B's construction is not adopted and nothing activates",
"note": "the coordinator's order (10:4x UK, 9 October 2026): Track B's reproduction findings as BOUND rows, NOT RUN, no pass or fail claimed for the package",
"cells": [
{
"cell": "theory:r15-05-track-b",
"cases": [
"R15-05"
],
"status": "NOT RUN",
"in_progress": true,
"method": "model",
"evidence": "docs/analysis/mhpow/b1/README.md; tools/mhpow/b1/fixtures/attacks-n10-k16.json; tools/mhpow/b1/fixtures/SHA256SUMS",
"note": "BOUND (B1, F-3): with ABH17's proven constant the paper's challenge count k exceeds N at lambda 256 and N 2^20 (to about n 25): no practical parameter in the proven regime. FINDING (B1, F-4): one labelling yields 193 to 200 accepted roots in 200 tries at 26 oracle calls per try, so a lottery over tau is not bound to the labelling cost (fixture A6, attacks-n10-k16.json, run on build-6). The package R15-05 (no improvised sampled-memory soundness) stays NOT RUN until the independent panel reads the reference and its verifier (17 of 17 named checks with known-failed tests)",
"claim_impact": "none"
}
],
"map_cell_requested": {
"theory:r15-05-track-b": {
"command": "the Track B cryptography read: docs/analysis/mhpow/b1 (the exact CPU reference of the MTP framework over DRSample, its verifier and the six malicious provers) by tools/mhpow/b1/run-b1.sh <out> all on a build box under a pid file, outputs byte-identical to tools/mhpow/b1/fixtures/SHA256SUMS",
"box_class": "build box, CPU only",
"fixtures": [
"tools/mhpow/b1/fixtures/"
],
"cases": [
"R15-05"
],
"coverage": {
"R15-05": "partial: the published primitive's reference, attacks and verifier refusals; the lottery binding (B2), the composed theorem (B4) and the panel's reading open"
}
}
}
}

61
tools/ci/test-map.json
View file

@ -822,6 +822,62 @@
"coverage": {
"TV-02": "partial: scheduled issuance on a straight chain and its reorg, two independent implementations EQUAL to the integer on devnet-4, mainnet and the TV-01 spec; paid, burned, locked, outstanding, DAG reds, the EVM ledger and duplicate records open; the independent panel has not read it"
}
},
"model:mhpow-b1-reference": {
"command": "tools/mhpow/b1/run-b1.sh <out_dir> all (a build box under its pid file; kat, attack and k-table files byte-identical to tools/mhpow/b1/fixtures/SHA256SUMS)",
"box_class": "build box, CPU only (build-6)",
"fixtures": [
"tools/mhpow/b1/fixtures/"
],
"cases": [
"POW-05",
"POW-06"
],
"coverage": {
"POW-05": "partial: the published primitive's attack fixtures only; the lottery binding and the composed accepted-proof theorem are B2/B4's",
"POW-06": "partial: verifier refusals and op counts in CPython; no compiled verifier, no admitted-rate timing"
}
},
"theory:r15-05-track-b": {
"command": "the Track B cryptography read: docs/analysis/mhpow/b1 (the exact CPU reference of the MTP framework over DRSample, its verifier and the six malicious provers) by tools/mhpow/b1/run-b1.sh <out> all on a build box under a pid file, outputs byte-identical to tools/mhpow/b1/fixtures/SHA256SUMS",
"box_class": "build box, CPU only",
"fixtures": [
"tools/mhpow/b1/fixtures/"
],
"cases": [
"R15-05"
],
"coverage": {
"R15-05": "partial: the published primitive's reference, attacks and verifier refusals; the lottery binding (B2), the composed theorem (B4) and the panel's reading open"
}
},
"class:int8": {
"command": "the int8-dominated research class: cut behind the research flag by the research lane at int8-50, int8-75, int8-90 and arx-16, arx-32 (igneum-pow export with the class string, on a build box); the hash lane's card rows in the P04 form (tools/1p5x/x1/gpu-pairs.sh, gpu-rows.py); the adversary lane's chip rows (tools/1p5x/x5/sweep.py, tools/1p5x/x8/mac.py); the census instruments (igneum-pow addrsite, x1cache)",
"box_class": "build box for the cut and the census; rented pods for the card rows; a build box for the chip model",
"fixtures": [
"F0"
],
"cases": [
"R15-03",
"R15-04"
],
"coverage": {
"R15-03": "partial: the held-out cache curves of each weight against the frozen control; the layer-8-off control is X1's row",
"R15-04": "partial: byte agreement and accepted-work savings of each weight's fold; X2 to X4's rows carry the rest"
}
},
"model:x7-memory-route": {
"command": "the X7 memory-route read: docs/analysis/class-v6/1p5x/x7 (the chip model's s on GDDR7, HBM3 and near-memory by tools/1p5x/x5/sweep.py on a build box under a pid file) against the hash lane's read-only twin row on a rented card at stock (the frozen object and its read-only twin, nvidia-smi beside it)",
"box_class": "a build box for the model; a rented card for the twin row",
"fixtures": [
"F0"
],
"cases": [
"R15-02"
],
"coverage": {
"R15-02": "partial: the memory route's f and s only; the component power and traffic model and its sensitivity are X0's and X5's rows"
}
}
},
"not_run": {
@ -831,7 +887,6 @@
"GPU-04": "the memory-clock ladder has no harness tonight: the project's own rig mines nothing under the the earlier devnet off order",
"GPU-06": "accepted work under ordinary connectivity needs the fault network F4",
"GPU-07": "the sustained thermal and power soak has no harness tonight: the project's own rig mines nothing under the the earlier devnet off order",
"POW-05": "amortised cheap winning attempts are the attack lanes' grind and era harnesses (tools/attack/f7-era, f9-grind), not in the release matrix; their rows come from those lanes",
"ROT-03": "miner-voted bring-forward needs a vote harness on the fault network F4",
"ROT-04": "seed-selection resistance is the census harness (the class v6 invention lane), not yet in the matrix",
"EVM-01": "no EVM conformance-vector harness is mapped tonight; the exec suite does not run the reference test vectors",
@ -995,10 +1050,6 @@
"TV-31": "TV-31 (token value gate, scope CORE): no gate executed; the harness is the owner role's (Qualified counsel); a supplemental layer, never counted as independent evidence; not yet named in the map",
"TV-32": "TV-32 (token value gate, scope LEADERSHIP CLAIM): no gate executed; the harness is the owner role's (Independent acceptance panel); a supplemental layer, never counted as independent evidence; not yet named in the map",
"R15-01": "R15-01 (1.5x programme package): no package executed; the harness is the lane's (X0); its batch names the case when it lands; not yet named in the map",
"R15-02": "R15-02 (1.5x programme package): no package executed; the harness is the lane's (X0, X5); its batch names the case when it lands; not yet named in the map",
"R15-03": "R15-03 (1.5x programme package): no package executed; the harness is the lane's (X1); its batch names the case when it lands; not yet named in the map",
"R15-04": "R15-04 (1.5x programme package): no package executed; the harness is the lane's (X2, X3, X4); its batch names the case when it lands; not yet named in the map",
"R15-05": "R15-05 (1.5x programme package): no package executed; the harness is the lane's (B1); its batch names the case when it lands; not yet named in the map",
"R15-06": "R15-06 (1.5x programme package): no package executed; the harness is the lane's (B3); its batch names the case when it lands; not yet named in the map",
"R15-07": "R15-07 (1.5x programme package): no package executed; the harness is the lane's (B2, B4); its batch names the case when it lands; not yet named in the map",
"R15-08": "R15-08 (1.5x programme package): no package executed; the harness is the lane's (X5, B5); its batch names the case when it lands; not yet named in the map",