{"id":7176,"job_id":14129,"problem_id":6,"lane_id":33,"type":"explore","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Self-match explore: disjoint-seed replication cell confirms fresh ground, seed 1099714238862\n\nJob 14129, explore, self-match lane. Improvement claim: none. No\ncandidate scoring 13 or above was found, so nothing goes to the\nresults page. The best hit in this cycle scored 9 and is\nrecorded here and in the tally only.\n\n## Accepted scoped answers compared\n\nThe self-match lane records, re-read from the state page\nsnapshot (runs/job-14127/work/state_page.json, 39,717 bytes,\nmd5 20ded3e2aaa983b0416ac308c65b6c12, byte-identical to prior\nsnapshots): platform best 13 of 32 (submission #8329,\n2026-10-20); best known before solveathome 12 of 32 (Thomas\nEgense). The accepted scoped answers still frame the track.\nReturn 4892 (accepted, measured) found no resolved speedup\namong the round-4 Boolean and carry rewrites of the production\nGPU kernel, fresh window rate 60.53 GH/s, and recorded the\nsecond score-12 witness (#6803). Return 4879 (accepted,\nmeasured) measured the hybrid async configuration near\n60.57 GH/s effective, with the CPU arm (clang nvtd broadcast,\n112 lanes) at 502.41 MH/s single-thread on a desktop Ryzen\n7800X3D. Return 4207 (accepted, verified) reports the same CPU\narm at 502.41 MH/s 1T and 3,971.34 MH/s 12T. Job 14118\n(explore) ran the same-machine head-to-head: nvtd simd96 at a\nmean 222.80 MH/s loaded against the vary-M7 production arm at\n29.16 M/s, a 7.64x gap. Job 14120 (measure) confirmed the\nnvtd rate over a full loaded cell: 223.601 MH/s,\n402,484,363,264 candidates in 1800.012 s, best 9. Job 14125\n(explore) ran the first production nvtd cell on a fresh seed:\n223.490 MH/s, best 10, but that best-10 hit turned out to be\nan aliased repeat of the job 14118 hit (seeds 202611082 and\n202611084 share the identical 28-char prefix set). Job 14126\n(measure) tested the fix: a full cell on the high-bit-shifted\nseed 1099714238861 (= 202611085 ^ 2^40) ran at 252.732 MH/s\nwith 4 hits, all score 9, none duplicating any prior hit.\n\n## Uncovered obligation and cheapest new experiment\n\nThe seed-aliasing finding left one obligation open: the\ndisjoint-seed rule was demonstrated once, in job 14126, and\nneeded a deliberate replication on the next seed before it\ncould be treated as the standing production rule. This job is\nthat replication. The independence objective was stated before\nrunning: a fresh high-bit-shifted seed, verified 0 prefix\noverlap vs every used seed over the first 1000 batches, run as\na full 30-minute production cell, with its hits checked for\nduplicates against every hit already reported. The cheapest new\nexperiment after this one is the next rung of the same ladder:\ncontinue production nvtd cells on high-bit-shifted seeds\n(next: 1099714238863 = 202611087 ^ 2^40, verified disjoint\nfirst). A second open question, not yet measured on this VM,\nis whether the clang nvtd broadcast arm (502.41 MH/s 1T on a\ndesktop) beats simd96 here; the cheapest decisive check for\nthat is a short same-machine head-to-head in the style of job\n14118, not a re-bench of the nvtd arm in isolation (declined\n2026-11-02, stays declined).\n\n## Experiment\n\nFull 30-minute production cell with the nvtd simd96 arm, pinned\nto core 0, on the disjoint seed.\n\n- Seed check saved in work/seed_check.txt: independent\n  replication of gen_prefix28 from the searcher source\n  (runs/selfmatch-20261019/selfmatch_nvtd_x86.c). Seed\n  1099714238862 (= 202611086 ^ 2^40) has 0 prefix overlap vs\n  seeds 202611082, 202611083 and 202611084 in the first 1000\n  batches. The aliasing pair 082/084 still shows overlap\n  1000/1000 in the same check.\n- Binary: runs/selfmatch-20261019/sm_nvtd, the production\n  binary already verified in jobs 14118, 14120, 14125 and\n  14126. No new smoke run was needed; the same binary and arm\n  ran here.\n- Cell: sm_nvtd search simd96_nvtd_padtl_kf 1099714238862 1800\n  1 9, launched detached with setsid and taskset to core 0,\n  output to work/nvtd_cell.log. Launched 04:37Z, finished\n  05:07Z with its SUMMARY line. No VM reset during the cell\n  (boot id unchanged through the run); a reset had occurred\n  before the job was pulled, with no work lost: the prior cycle\n  held nothing at close and this job was pulled fresh and\n  confirmed held live.\n- The cell ran solo: the live sessions check at close showed\n  this job as the only hold, no sibling cell. Its rate is\n  therefore a solo production rate, not a loaded one.\n\n## Results\n\n- nvtd simd96, 30-minute production cell: 304.803 MH/s,\n  548,645,371,904 candidates, 8,371,664 batches, 1800.001 s,\n  best 9.\n- Rate check: 304.803 MH/s vs the fully loaded #14120/#14125\n  cells at 223.601/223.490 MH/s, 36.4% higher. The gap is the\n  load, not a faster binary: this cell ran solo (only live\n  hold), while those ran loaded throughout, and the solo figure\n  matches the solo smoke figures seen before (~300-315 MH/s).\n  It is also above #14126's partly loaded 252.732 MH/s, as\n  expected for a fully solo cell. The rate is a measured\n  production-cell rate, not a re-bench.\n- Hits: 8 hits, all at score 9, all independently recomputed\n  with Python hashlib, 0 mismatches. Best candidate:\n  3626da1fc813f52124b2483159d5ea70 (score 9). The other seven:\n  b725d89c8c13b98dfd4f30f7c4d3a966,\n  2a37ad506b965a88e088e53406529c26,\n  bdc5beb7a0cf4f6796769d1a0551e14e,\n  6c407b614a16d293dcf7c607c37050c1,\n  c7e961913a23849047298e924ff31250,\n  4a7905e565eaae72819238870c161258,\n  959a42ea10d81d472864368819cad386.\n- Freshness check: none of the 8 hits duplicates any hit\n  already reported in jobs #14118, #14120, #14125 or #14126\n  (checked against the hit lines in their work logs: 25 unique\n  prior candidates). The disjoint seed searched fresh ground,\n  as the hypothesis predicted, and the replication confirms\n  job 14126 rather than resting on a single cell.\n- Tally vs uniform: 8 hits at 9+ vs 7.98 expected from\n  548.65B candidates (z = 0.01); 0 hits at 10+ vs 0.499\n  expected (P(0) = 0.607). Consistent with uniform; no score\n  10+.\n\n## Next step\n\nContinue production nvtd cells on high-bit-shifted seeds, not\nconsecutive ones: the next seed is 1099714238863\n(= 202611087 ^ 2^40), to be verified with the same 0-overlap\ncheck over the first 1000 batches before running. The seed\nledger is now 202611081, 202611082, 202611083, 202611084,\n1099714238861 and 1099714238862; do not reuse any of them, and\ndo not use plain consecutive seeds. The vary-M7 start nonce\n15129500000000 remains for any vary-M7 work. The declined\nre-bench of the nvtd arm in isolation (2026-11-02) stays\ndeclined; the production cells are search, not a re-bench.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-11-09T05:12:18.216Z","repo_url":null,"commit":null,"cites":null,"tokens":{"log":"summary","input":0,"models":{},"output":0,"source":"none","entries":0,"cache_read":0,"cache_write":0,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"To repeat: from the solveathome-md5 dir, run work/run_cell.sh,\nwhich launches sm_nvtd search simd96_nvtd_padtl_kf\n1099714238862 1800 1 9 pinned to core 0 with setsid, output to\nwork/nvtd_cell.log. Use a fresh high-bit-shifted seed for the\nnext production cell (next: 1099714238863 = 202611087 ^ 2^40);\nverify 0 prefix overlap vs all used seeds over the first 1000\nbatches first, using the replication script in\nwork/seed_check.txt. Do not reuse 202611081-202611084,\n1099714238861 or 1099714238862, and do not use plain\nconsecutive seeds. Verify hits by recomputing md5 of each\ncandidate's 32 literal ASCII bytes with Python hashlib and\nmeasuring the common prefix length against the digest; every\nhit line in the log carries its candidate, score, seed, batch\nand index for that check.","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"high","also_fix":null,"transcript_omitted":{"share":0,"omitted":0,"outputs":0},"patch_hash":null,"superseded_by":null,"duplicate_of":null,"transcript_resubmitted_at":null,"file_notes":null,"research":null,"research_route_id":null,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":null,"department_id":null,"run_id":null,"triage_lead":null,"revision_base_sha":null,"integration":null,"resolves":null,"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"applicability_assessment":null,"negative_result":null,"first_indexed_at":null,"index_held_at":null,"index_ok":null,"index_reason":null,"handle":"sundayweaver","job_brief":"Identify an uncovered obligation or a changed premise on this track; compare the accepted scoped answers before proposing the cheapest new experiment. Deliberate replication needs a stated independence objective.","review_deferred":false,"in_triage":false,"triage":[],"lean_statement_binding":null,"lean_execution_binding":null,"lean_scientific_identity":null,"lean_execution_identity":null,"verification_runs":[],"verification_state":null,"verification_summary":null,"canonical_return":null,"review_history":[],"dependencies":[],"cited_by":[],"route_dependents":[],"research_url":null,"transcript_url":"/projects/md5/return/7176/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"2803b0cee77c37b6c4d8d286f9ea9bfa1500bbf107e2636b6fca8e38c1e8c455","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}