{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AHKZBBHQZM2VMI4MQ6ZQBKILDF","short_pith_number":"pith:AHKZBBHQ","schema_version":"1.0","canonical_sha256":"01d59084f0cb3556238c87b300a90b194fb93bdf96023362a2cc26cd731a6954","source":{"kind":"arxiv","id":"2606.18463","version":1},"attestation_state":"computed","paper":{"title":"Mixed-Precision Communication-Avoiding SGD for Generalized Linear Models on GPUs","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG","cs.NA","math.NA","stat.ML"],"primary_cat":"cs.DC","authors_text":"Aditya Devarakonda, Giulia Guidi, Irene Sim\\'o Mu\\~noz","submitted_at":"2026-06-16T20:14:34Z","abstract_excerpt":"Distributed stochastic gradient descent (SGD) is limited by communication rather than computation, since each iteration requires an AllReduce across processes. Communication-avoiding SGD (CA-SGD) amortizes communication over $s$ iterations by replacing $s$ consecutive AllReduces with a single AllReduce of an $sb\\times sb$ Gram matrix, trading more computation and bandwidth for fewer synchronization points. Modern GPUs with matrix hardware and reduced-precision formats offset this by accelerating the Gram GEMM and shrinking BF16 traffic. We study mixed-precision CA-SGD for generalized linear mo"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.18463","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.DC","submitted_at":"2026-06-16T20:14:34Z","cross_cats_sorted":["cs.LG","cs.NA","math.NA","stat.ML"],"title_canon_sha256":"88df11e01aecbdb82f9b3df2d5722abb69b933513a81325d64ccd7b80e8075a0","abstract_canon_sha256":"5b63d1658991f88bf61a2ed342cb94b02d3eda5739cd9d27c6d2bcbdad1c5512"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:11:01.805147Z","signature_b64":"49auZEoOpgiGUM4725XKv42WdfIHsXf/VtDwc2zehmPT2CKZ2vZXutmVsR5uGZIgK9Rme6mddre73/hSC2cHDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"01d59084f0cb3556238c87b300a90b194fb93bdf96023362a2cc26cd731a6954","last_reissued_at":"2026-06-19T16:11:01.804777Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:11:01.804777Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mixed-Precision Communication-Avoiding SGD for Generalized Linear Models on GPUs","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG","cs.NA","math.NA","stat.ML"],"primary_cat":"cs.DC","authors_text":"Aditya Devarakonda, Giulia Guidi, Irene Sim\\'o Mu\\~noz","submitted_at":"2026-06-16T20:14:34Z","abstract_excerpt":"Distributed stochastic gradient descent (SGD) is limited by communication rather than computation, since each iteration requires an AllReduce across processes. Communication-avoiding SGD (CA-SGD) amortizes communication over $s$ iterations by replacing $s$ consecutive AllReduces with a single AllReduce of an $sb\\times sb$ Gram matrix, trading more computation and bandwidth for fewer synchronization points. Modern GPUs with matrix hardware and reduced-precision formats offset this by accelerating the Gram GEMM and shrinking BF16 traffic. We study mixed-precision CA-SGD for generalized linear mo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.18463","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.18463/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.18463","created_at":"2026-06-19T16:11:01.804838+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.18463v1","created_at":"2026-06-19T16:11:01.804838+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.18463","created_at":"2026-06-19T16:11:01.804838+00:00"},{"alias_kind":"pith_short_12","alias_value":"AHKZBBHQZM2V","created_at":"2026-06-19T16:11:01.804838+00:00"},{"alias_kind":"pith_short_16","alias_value":"AHKZBBHQZM2VMI4M","created_at":"2026-06-19T16:11:01.804838+00:00"},{"alias_kind":"pith_short_8","alias_value":"AHKZBBHQ","created_at":"2026-06-19T16:11:01.804838+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF","json":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF.json","graph_json":"https://pith.science/api/pith-number/AHKZBBHQZM2VMI4MQ6ZQBKILDF/graph.json","events_json":"https://pith.science/api/pith-number/AHKZBBHQZM2VMI4MQ6ZQBKILDF/events.json","paper":"https://pith.science/paper/AHKZBBHQ"},"agent_actions":{"view_html":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF","download_json":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF.json","view_paper":"https://pith.science/paper/AHKZBBHQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.18463&json=true","fetch_graph":"https://pith.science/api/pith-number/AHKZBBHQZM2VMI4MQ6ZQBKILDF/graph.json","fetch_events":"https://pith.science/api/pith-number/AHKZBBHQZM2VMI4MQ6ZQBKILDF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF/action/storage_attestation","attest_author":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF/action/author_attestation","sign_citation":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF/action/citation_signature","submit_replication":"https://pith.science/pith/AHKZBBHQZM2VMI4MQ6ZQBKILDF/action/replication_record"}},"created_at":"2026-06-19T16:11:01.804838+00:00","updated_at":"2026-06-19T16:11:01.804838+00:00"}