{"id":6706,"job_id":13143,"problem_id":6,"lane_id":33,"type":"measure","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Job #13143 measure report: repair of return #4850 to a checkable state\n\n## What this job does\nReturn #4850 (measure, job 9706, status rejected) could not be verified with what it supplied. This job brings the parts that can be checked to a checkable state, and states plainly the parts that cannot. It does not redo the original campaign from scratch, and it does not claim a new search recovered historical observations.\n\n## Duplicate campaign, explained\nReturn #4851 (measure, job 9708, status accepted) carries a report and recipe verbatim identical to #4850. #4851 is the accepted line for this campaign; #4850 is the rejected duplicate. This return attaches verification once, on the #4850 credit line as the brief directs, not as a second duplicate return.\n\n## Verified here, on this host\nBuild: gcc -O3 -march=znver4 -mtune=znver4 -o sm_nvtd selfmatch_nvtd_x86.c -lpthread, using #3959's file selfmatch_nvtd_x86.c, sha256 6eef18a32be88476938b6aa0bde3863c3d22b0596d58ee5ef2b17dd2e0469ae2, which matches the hash recorded on #3959.\nSelftest: ./sm_nvtd selftest 802401 16 -> SELFTEST PASS, all 18 arms agree (best=5, batch=14, j=7819).\nBenches, ./sm_nvtd bench <arm> 99991 30000 1, 1 thread, 1,966,080,000 candidates each:\n- simd96_nvtd_padtl_kf: 314.0 MH/s (6.26 s)\n- simd64_kf: 235.2 MH/s (8.36 s)\n- simd64_nvtd_padtl_kf: 231.7 MH/s (8.49 s)\n#4850 claimed 305.2, 224.1 and 225.0 MH/s for the same three arms. The ranking is reproduced exactly, simd96 remains the fastest x86 arm, at 1.33x simd64_kf and 1.36x simd64_nvtd_padtl_kf here, and the rates are close to #4850's, within host-clock variation. The transfer hypothesis in #4850 survives this check.\nScore-10 witness, independently rehashed: printf '%s' f71e3fa37c65738821381652a4db9f09 | md5sum -> f71e3fa37c242ecb7d84398d34e7a679, first 10 nibbles match, score 10. This witness is verified.\n\n## Corrected null statement\nWith N = 2.917e12 as claimed in #4850:\n- E[==9] = N*15/16^10 = 39.8; E[==10] = 2.49; E[==11] = 0.16\n- E[>=9] = N/16^9 = 42.4; E[>=10] = 2.65; E[>=11] = 0.17\n#4850 quoted E[9] = 42 and E[10] = 2.6. Those are the threshold (>=) expectations, not exact-score expectations, and are corrected here.\nPoisson tails, conditional on N: observed exact-9 count 57 at mean 39.8 gives P(X>=57) = 0.0060. Observed >=9 count 58 at mean 42.4 gives P(X>=58) = 0.013. Observed exact-10 count 1 at mean 2.49 gives P(X<=1) = 0.29, unremarkable. The exact-9 tally is higher than the corrected null mean, but a single campaign tally at p ~ 0.006 is not, on its own, evidence of structure, and it cannot be rechecked without the hit list below.\n\n## CPU accounting\n47 chunks x 300 s = 14,100 CPU-s = 3.92 CPU-h, matching the brief's about 3.9. This is derived from the claimed chunk count and duration, not independently verified, because the chunk logs were not supplied.\n\n## Not recoverable, and not claimed otherwise\n- The 47 per-chunk logs (seed, candidates tried, rate per chunk) were not in #4850 or #4851 as fetched, so N cannot be summed from logs here. A new search cannot recover those historical observations.\n- The full deduplicated list of 57 score-9 hits (input string and score) was not supplied, so the 1 + 57 count cannot be rehashed in full. Only the score-10 witness above is independently rehashed here.\n- A cited witness #6615 was not found on the server as fetched (404, no such return), so it is not used as evidence for anything in this return.\n- No candidate at score 13 or above was produced or claimed here, so nothing is posted to the results page.\n\n## Rung defended\nBench, selftest, source hash, and the score-10 witness are verified. The campaign tally, null arithmetic conditional on N, and CPU accounting are reproducible from the claimed N and chunk count, with the corrections above. The original campaign's per-chunk and per-hit records remain unverified, because they were not supplied and a new search cannot recreate them.\n","patch":null,"cpu_hours":3.92,"hashes":{},"author_rung":null,"status":"pending","final_rung":null,"created_at":"2026-10-31T03:56:21.854Z","repo_url":null,"commit":null,"cites":{"returns":[4850,3959]},"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":"Build: gcc -O3 -march=znver4 -mtune=znver4 -o sm_nvtd selfmatch_nvtd_x86.c -lpthread, where selfmatch_nvtd_x86.c is #3959's file, sha256 6eef18a32be88476938b6aa0bde3863c3d22b0596d58ee5ef2b17dd2e0469ae2.\nSelftest: ./sm_nvtd selftest 802401 16 -> expect SELFTEST PASS, all 18 arms agree, exit 0 (about 2 s).\nBench: ./sm_nvtd bench simd96_nvtd_padtl_kf 99991 30000 1 -> expect about 314 MH/s; ./sm_nvtd bench simd64_kf 99991 30000 1 -> about 235 MH/s; ./sm_nvtd bench simd64_nvtd_padtl_kf 99991 30000 1 -> about 232 MH/s; each about 6-9 s, host-clock dependent, ranking simd96 first.\nScore-10 witness: printf '%s' f71e3fa37c65738821381652a4db9f09 | md5sum -> f71e3fa37c242ecb7d84398d34e7a679 (first 10 nibbles match, score 10).\nNull check: with N = 2.917e12, E[==9] = N*15/16^10 = 39.8, E[>=9] = N/16^9 = 42.4; Poisson P(X>=57) at mean 39.8 is 0.0060.\nCampaign rerun (for a new campaign only, not a recovery of the old one): ./sm_nvtd search simd96_nvtd_padtl_kf <seed> 300 1 9 with distinct 64-bit seeds per process; expect about 200 MH/s per worker under 2-core load and score-9 hits at about 16^-9 per candidate. Checking cost for the verified parts: build about 5 s, selftest about 2 s, benches about 25 s total, md5sum instant.","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":"2026-10-31T03:56:21.854Z","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":"Return #4850 (measure) could not be verified with what it supplied. It is not wrong as far as anyone knows; it is not checkable yet. Your job is to bring it to a checkable state, not to redo it from scratch.\n\nWhat the reviewer said a checkable return needs:\n> Attach once, on #4851's line rather than as a new duplicate return:\n> (1) The output of `./sm_nvtd selftest 802401 16` and of `./sm_nvtd bench <kernel> 99991 30000 1` for simd96_nvtd_padtl_kf, simd64_kf and simd64_nvtd_padtl_kf on the EPYC 9D25 host (gcc 13.3, -march=znver4), with the build line and the source sha256 of #3959's file.\n> (2) The 47 chunk logs of `search simd96_nvtd_padtl_kf <seed> 300 1 9`, giving the seed, the candidates tried and the rate per chunk, so that N can be summed.\n> (3) The full list of score >= 9 hits (input string and score), deduplicated, so that the 1 + 57 count can be rehashed.\n> (4) A corrected null statement: E[==9] = N*15/16^10, with a Poisson tail for the observed count.\n> (5) Structured cites for #3959 and cpu_hours of about 3.9.\n\n> Preserve the original same-host benchmark/selftest outputs with all tested arms, repetitions and compiler details; exact per-chunk seeds/counts/durations and partial work; complete deduplicated hit records; actual scientific CPU accounting; and reconciliation of exact-score versus threshold counts and null-model language. Explain the duplicate campaign in #4851. Existing verified witness #6615 does not validate the missing campaign or timing observations. A new search cannot recover those historical observations.\n\nStart from the original: its report and files are at <project base>/return/4850. Reproduce what it claims with the cheapest credible recipe a stranger can run without repeating discovery: exact commands, served script paths, inputs, expected outputs and their sha256, run time. Where the original claim does not survive, say so at the rung you can defend.\n\nReturn as this job with `\"recipe_md\"` filled in and `\"cites\": { \"returns\": [4850] }` so the original author is credited on acceptance.\n\n---\n\nOriginal assignment:\n\nStudy how a candidate's 32 ASCII bytes flow through the 64 steps into the first digest characters, and use what you learn to reach a longer matching prefix. Ideas to test: which message words the first output word depends on most, fixing a prefix and solving for the rest, early-exit tests on the first output word, meet-in-the-middle on the step function. Start from the algorithm, not the search. Read the track's state of the art in this brief (the fastest known way per machine, what worked, the dead ends) and research/QUESTIONS.md, then state one hypothesis about MD5's structure that would make this track cheaper than generic search, and why you expect it. Test it with the smallest experiment that could refute it, against a measured baseline on the same machine. Submit the best candidates the experiment produced. The report is a finding: the hypothesis, the experiment, what it showed about MD5 (positive or negative, with numbers), and what the next run should try. Send what you learned as a state_entry with your return (the method, the exact hardware, the measured throughput, the best reached), so the track's state of the art carries it. If the run used only a known tool or plain search, report it as a baseline measurement.","review_deferred":false,"in_triage":true,"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/6706/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"5cce41b2512ebf99614ca4ef7184cf749de0beb24ecc244d1845b722aa26cb4b","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"pending","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}