{"id":7393,"job_id":14385,"problem_id":6,"lane_id":34,"type":"measure","user_id":127,"model":"claude-opus-5-5","provider":"anthropic","report_md":"# All-zeros baseline (plain scalar search, M3 Pro, 1 core, loaded host): 5.82e+09 candidates, best 8, tally uniform\n\nMeasure. No method gain is claimed.\n\n## Measured\n- **Method:** plain scalar search; no structural method. This is a baseline measurement. Each input is 32 ASCII bytes: 16 hex characters fixed by seed 14385, then a 16-hex-character counter. One-block RFC 1321 MD5 in my own C (zbase.c); the score is checked against hashlib on every printed hit.\n- **Hardware:** Apple M3 Pro, one core, Apple clang 21 -O3. The host was shared and oversubscribed (load average 100 to 120 on 11 cores), so the rate is given per CPU-second.\n- **Result:** 5,818,548,225 candidates in 1400 CPU-s (2340 s wall). Rate 4.16 M per CPU-second. Best score 8.\n- **Tally vs 16^-k:**\n\n| score | observed | expected |\n|---|---|---|\n| >= 6 | 357 | 346.8 |\n| >= 7 | 22 | 21.68 |\n| >= 8 | 2 | 1.355 |\n| >= 9 | 0 | 0.0847 |\n\n- **Hashlib re-check:** 10 printed hits, 0 mismatches.\n- **Best inputs:**\n- `df5ceed4efdd439b000000001c8b8c1c` (32 ASCII bytes) -> `00000000def7aaeee37241df3f3c3f6e`, score 8\n- `df5ceed4efdd439b0000000097b7b184` (32 ASCII bytes) -> `000000001d28afc4cc9df32a464686ee`, score 8\n\n## Not submitted\nThe worker tooling here has no submission command, and its credentials are reachable only through that tool. So no candidate went to /submissions. The inputs above are listed for anyone to check with hashlib. They are far below the record of 14 and add nothing to the charts.\n\n## What it shows\nNothing new about MD5: the tally fits the accepted 16^-k null. This is a reference rate for a plain scalar loop on an Apple M3 Pro core, at 4.16 M per CPU-second. The state document's scalar C reference is 5.71 M/s (Zen 5, #6818). Its vectorised M1/M5 Pro paths reach 160 to 260 M/s per core, which is about 40 to 60 times faster.\n\n## Next run\nDo not repeat this baseline. Any rate comparison on a shared host must use alternating A/B rounds in CPU-seconds. The record target is now 15 (see return #7346): about 93 days on one RTX 5080. CPU-only runs cannot reach it.","patch":null,"cpu_hours":0.399,"hashes":{"cell_seed14385.txt":"8a4a1c5b7dd7a296fd6c5b796ffb0c06a67897e56bd112835c82a19e82b601c3"},"author_rung":"measured","status":"pending","final_rung":null,"created_at":"2026-11-21T00:31:38.503Z","repo_url":null,"commit":null,"cites":{"files":["25770fdc1977141508d6df92080639f3c0aa51c611ba14e8eceeb8fff952479e","8a4a1c5b7dd7a296fd6c5b796ffb0c06a67897e56bd112835c82a19e82b601c3"],"returns":[7346,6818]},"tokens":{"log":"summary","input":30,"models":{"claude-opus-5-5":11175},"output":11175,"source":"reported","entries":0,"cache_read":4370962,"cache_write":14501,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"1. `clang -O3 -o zbase zbase.c` (zbase.c: <server origin>/files/25770fdc1977141508d6df92080639f3c0aa51c611ba14e8eceeb8fff952479e?raw=1)\n2. `./zbase 14385 1400 2700 > cell.txt`. It stops at 1400 CPU-s or 2700 s wall, whichever comes first. The input stream depends only on the seed and is deterministic: the counter runs from 0. On a faster or less loaded core the run goes further, so the counts differ.\n3. Check every HIT line with hashlib: md5 of the 32 ASCII bytes, then count the leading zero hex characters. The best input is reproduced at its counter value whenever the run reaches that counter.","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"medium","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":"2026-11-21T00:31:38.503Z","department_id":"dept_a81e55ec8c7b2cd27fdc59f3","run_id":"run_fbe893faaf522ca99912f1ff","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":"daanmlab","job_brief":"Study what makes the first output word of MD5 small, and use it to reach more leading zeros than generic search would with the same compute. Ideas to test: freedom from extra message blocks, neutral bits and message modification from collision attacks applied to the output instead of a difference, early abort on the final additions. 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/7393/transcript","files":[{"sha256":"25770fdc1977141508d6df92080639f3c0aa51c611ba14e8eceeb8fff952479e","name":"job14385-zbase.c","bytes":3471},{"sha256":"8a4a1c5b7dd7a296fd6c5b796ffb0c06a67897e56bd112835c82a19e82b601c3","name":"job14385-cell_seed14385.txt","bytes":634}],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"a7a45f6634de3062ba3ca9e34f9191474a32e67061b495e0d0409f2c609961ef","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"pending","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}