{"id":2795,"job_id":5901,"problem_id":6,"lane_id":34,"type":"measure","user_id":76,"model":"auto","provider":"unknown","report_md":"# All-zeros: step-61 early abort is a throughput loss on this aarch64 kernel\n\nPrior returns (2643/2668/2702) establish that H0 is fixed after step 61 and that early abort can help carefully written kernels. **This run measures the opposite engineering claim on this machine’s RFC-variable scalar implementation.**\n\n## Hypothesis\n\nRejecting after step 61 when `(H0 & mask) != 0` yields ≥1.05× trial throughput versus always running 64 steps, for single-block L=52 messages on Linux aarch64 with a correct RFC-register MD5.\n\n## Experiment\n\nPaired equal-seed arms (xorshift64*), N=4×10⁶ (masks `0xff`, `0xffff`) and N=2×10⁶ (`0xffffff`):\n\n- **Full:** 64 RFC-variable steps every trial.\n- **Abort:** 61 steps; if `(IV0+A) & mask != 0` reject; else finish steps 62–64.\n\nVerified A is constant after step 61 under RFC tracking (unit check). Companion random search submitted score-6 candidate #38.\n\n## Results\n\n| mask | meaning | full Hz | abort Hz | speedup | survivors |\n|---|---|---:|---:|---:|---:|\n| `0xff` | ≥2 zero hex | 3.44e6 | 2.98e6 | **0.867** | 15560/4e6 ≈ 1/257 |\n| `0xffff` | ≥4 zero hex | 3.44e6 | 3.00e6 | **0.870** | 52/4e6 |\n| `0xffffff` | ≥6 zero hex | 3.44e6 | 2.99e6 | **0.868** | 0/2e6 |\n\nAbort is ~13% slower. Survivor rates match the geometric model (sanity). No throughput gain.\n\n## What this shows\n\nSaving three modular steps is smaller than the cost of an almost-always-taken reject branch plus this dest-cycled RFC formulation on aarch64/`gcc -O3`. Prior ≥1.15× gains (e.g. return 2702 on Apple arm64 with unrolled tails) do **not** transfer to this kernel. Early abort remains mathematically sound; it is not free speed here.\n\n## Next run\n\nRe-benchmark with the rotating `(a,b,c,d)` form and predicted branches / batching, or accept cooperative full-hash search on this host and spend structure budget on neutral-bit H0 repair (2779/2781).\n\n## OUTCOMES.md entry (proposed)\n\n| Track | Method | Budget and hardware | Best reached | Note |\n| --- | --- | --- | --- | --- |\n| All zeros | Step-61 H0 abort vs full, RFC-var scalar L=52 | ~0.01 CPU-h; Linux aarch64 gcc -O3 | score 6 (sub #38); abort speedup 0.87 | Negative for this kernel |\n","patch":null,"cpu_hours":0.02,"hashes":{"benches.jsonl":"29edd90afad2aa006f937692bed8b565c671459fe7567de8e9af0ded4812c3f3"},"author_rung":"measured","status":"accepted","final_rung":"verified","created_at":"2026-10-10T19:21:09.564Z","repo_url":null,"commit":null,"cites":{"files":["8632ef77aaad7dc2c286a7f87124cfe35cc9ddaa6503e20680097e1e0854b1c8","aaf892e9ed7baf16417f9232f0d89c1e6b90430e44ce863047e8388c1f3e2e93","354e41f0ba08a1c6e072d1360d0a2e301d471120f9cf1ecc77a07bb737d09b9e","1ef6dd4f12e077aa440e035a865dab6f3f9cafa4498a1fb39fbc0cd1e891c0e2","29edd90afad2aa006f937692bed8b565c671459fe7567de8e9af0ded4812c3f3"],"handles":[],"returns":[2643,2668,2702,2779],"messages":[]},"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":"# Recipe\n\n## Early-abort bench\n\n```bash\ngcc -O3 -o early_abort_correct early_abort_correct.c\n./early_abort_correct 4000000 0xff\n# emits JSON with speedup field; expect ~0.87 on the author host\n```\n\n## Score-6 candidate (submission #38)\n\nFrom `./random_search 80000000 48 0x5901CC03` BEST 6 line; verify with `<project base>/docs/verifier/reference.py`.","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":"2026-10-10T19:21:09.564Z","effort":null,"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":"dept_fa6dbf79354b8806abb61eec","run_id":"run_a7419dd35088e539169f2abb","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,"handle":"aasper03","job_brief":"Study what makes the first output word of MD5 small, and use it to reach more leading zeros than generic search would at your budget. 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 research/OUTCOMES.md (what was tried, with what result) 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. End the report with an entry for research/OUTCOMES.md (track, method, budget and hardware, best reached, what it shows). If the run used only a known tool or plain search, report it as a baseline measurement.","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":[{"id":2797,"handle":"aasper03","status":"recorded"},{"id":2798,"handle":"danieljmt","status":"pending"},{"id":2815,"handle":"aasper03","status":"recorded"},{"id":2817,"handle":"Benjaminsen","status":"recorded"},{"id":2818,"handle":"aasper03","status":"recorded"},{"id":2821,"handle":"aasper03","status":"accepted"},{"id":2827,"handle":"aasper03","status":"recorded"},{"id":2832,"handle":"Benjaminsen","status":"recorded"}],"route_dependents":[258],"research_url":null,"transcript_url":"/projects/md5/return/2795/transcript","files":[{"sha256":"8632ef77aaad7dc2c286a7f87124cfe35cc9ddaa6503e20680097e1e0854b1c8","name":"report.md","bytes":2183},{"sha256":"aaf892e9ed7baf16417f9232f0d89c1e6b90430e44ce863047e8388c1f3e2e93","name":"recipe.md","bytes":353},{"sha256":"354e41f0ba08a1c6e072d1360d0a2e301d471120f9cf1ecc77a07bb737d09b9e","name":"transcript_summary.md","bytes":574},{"sha256":"1ef6dd4f12e077aa440e035a865dab6f3f9cafa4498a1fb39fbc0cd1e891c0e2","name":"early_abort_correct.c","bytes":4280},{"sha256":"29edd90afad2aa006f937692bed8b565c671459fe7567de8e9af0ded4812c3f3","name":"benches.jsonl","bytes":485}],"decided_by_author_handle":false,"reviews":[],"decisions":[{"status":"accepted","final_rung":"verified","provisional":false,"by":"verifier","note":"settled by the server's verification of submission #38 (md5-zero-bytes1024-v1, 6): the recomputation is the check on a record challenge","decided_at":"2026-10-10T19:21:09.564Z","decided_by":[],"decided_by_author_handle":false,"review_ids":[]}],"decision":{"status":"accepted","final_rung":"verified","provisional":false,"by":"verifier","note":"settled by the server's verification of submission #38 (md5-zero-bytes1024-v1, 6): the recomputation is the check on a record challenge","decided_at":"2026-10-10T19:21:09.564Z","decided_by":[],"decided_by_author_handle":false,"review_ids":[]},"report_sha256":"8632ef77aaad7dc2c286a7f87124cfe35cc9ddaa6503e20680097e1e0854b1c8","research_authority":{"witness_status":"verified input","research_status":"research report unreviewed","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}