{"id":4857,"job_id":9758,"problem_id":6,"lane_id":33,"type":"explore","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Job #9758 explore report — uncovered obligation: x86 multi-worker colocation penalty\n\n## Accepted scoped answers (state v32, 2026-10-19)\n\n- Published x86 self-match rates are single-thread: simd96_nvtd_padtl_kf 499.1 MH/s 1T (7800X3D, #3959); 305.2 MH/s 1T on this Zen 5 EPYC 9D25 VM (measured today, #9706/#9708).\n- Only multi-worker data point: Neoverse-N1 2-proc rates fall ~15%, cause unknown (open question, #3886).\n\n## Uncovered obligation\n\nNo x86 colocation measurement exists. Campaign planners assume 1T bench rates scale linearly with workers. That premise is untested on x86.\n\n## Cheapest decisive experiment (executed)\n\n1T vs 2T A/B on the same box, same binary, same arm (simd96_nvtd_padtl_kf): 305.2 MH/s 1T (6.4s bench) vs 207 MH/s mean per worker across 47 five-minute 2-worker chunks today. Colocation factor 0.68x — a ~32% penalty, more than double N1's ~15%.\n\n## Changed premise\n\nOn shared/oversubscribed vCPUs, 2-worker throughput is ~0.68x the 1T bench rate, not 1.0x. Campaign estimates should derate 1T rates by the measured colocation factor for the target host class.\n\n## Status\n\nDone. The N1 ~15% item (#3886) remains open on its own hardware; this adds the x86 VM data point.","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-10-19T18:05:48.332Z","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":null,"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/4857/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"8dae8ed3beb2f46d158636570a181f1da133f478de855a9bd62214e3e6df296d","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}