{"paper":{"title":"Correctly Rounded Functions For Vector Applications: A Performance Study","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"SIMD algorithms for correctly rounded single-precision functions form the core of a new vector math library planned for mid-2026.","cross_cats":[],"primary_cat":"cs.MS","authors_text":"Andrey Stepin, Cristina Anderson, Marius Cornea, Mihai Tudor Panu","submitted_at":"2026-05-15T02:39:39Z","abstract_excerpt":"Following recent interest in correctly rounded math library functions (as currently recommended by the IEEE 754 standard), we have designed several SIMD algorithms for one-input single precision functions and integrated them into our CPU math library; these will form the core of the first correctly rounded vector math library, to be available to users in mid-2026. To take advantage of the cross-platform bitwise reproducibility afforded by correct rounding, we adapted and evaluated a few SIMD implementations on graphics processing units (GPU). In addition, we designed and evaluated proof-of-con"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we have designed several SIMD algorithms for one-input single precision functions and integrated them into our CPU math library; these will form the core of the first correctly rounded vector math library, to be available to users in mid-2026","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The designed SIMD algorithms achieve correct rounding (per IEEE 754) while delivering competitive performance on vector hardware; this premise is stated in the abstract but not supported by any verification details or data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Design and performance study of SIMD algorithms for correctly rounded one-input math functions in vector CPU and GPU applications.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"SIMD algorithms for correctly rounded single-precision functions form the core of a new vector math library planned for mid-2026.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"7b9f073e41d2fd35fcf89cf020e43fd90f8eb959f562022be9ce1e8a08fe1dcd"},"source":{"id":"2605.15547","kind":"arxiv","version":1},"verdict":{"id":"dd4c27b5-466b-4e47-80c9-1250734b4971","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T18:27:36.265007Z","strongest_claim":"we have designed several SIMD algorithms for one-input single precision functions and integrated them into our CPU math library; these will form the core of the first correctly rounded vector math library, to be available to users in mid-2026","one_line_summary":"Design and performance study of SIMD algorithms for correctly rounded one-input math functions in vector CPU and GPU applications.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The designed SIMD algorithms achieve correct rounding (per IEEE 754) while delivering competitive performance on vector hardware; this premise is stated in the abstract but not supported by any verification details or data.","pith_extraction_headline":"SIMD algorithms for correctly rounded single-precision functions form the core of a new vector math library planned for mid-2026."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15547/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T19:34:36.569043Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T19:01:19.033481Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:41:27.474830Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T17:41:56.098420Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"a1ac3de4e14220def38cbea7905a67f172a8de7a80e7e32b051e797786149f09"},"references":{"count":14,"sample":[{"doi":"","year":1991,"title":"Fast evaluation of elementary mathematical functions with correctly rounded last bit","work_id":"2909bfaa-3067-4dd6-9863-4dfca58bf75f","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2010,"title":"Handbook of Floating-Point Arithmetic","work_id":"5b21a748-2748-4928-aa09-d66dbadaedc6","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Correctly Rounded Math Libraries without Worrying about the Application’s Rounding Mode","work_id":"248bc5cb-6791-4d7b-8ec2-d82375873612","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2026,"title":"Correctly rounded evaluation of a function: why, how, and at what cost?","work_id":"d57e3120-369f-4912-9c79-af770b1c0b93","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"The CORE-MATH Project","work_id":"48a04e78-0014-4dd2-9159-1eda0ee103df","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":14,"snapshot_sha256":"39169c3edf9a37477837ad426c0c55858c5458aa1fba0321455e49d84adf7653","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}