{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:7N6MR4HH2V2MKAYWZNSCIXTWL2","short_pith_number":"pith:7N6MR4HH","schema_version":"1.0","canonical_sha256":"fb7cc8f0e7d574c50316cb64245e765e9eb550c2bf32826d4fb60dae529da195","source":{"kind":"arxiv","id":"2202.00433","version":3},"attestation_state":"computed","paper":{"title":"TopoOpt: Co-optimizing Network Topology and Parallelization Strategy for Distributed Training Jobs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.NI","authors_text":"Anthony Kewitsch, Dheevatsa Mudigere, Manya Ghobadi, Moein Khazraee, Weiyang Wang, Ying Zhang, Zhihao Jia, Zhizhen Zhong","submitted_at":"2022-02-01T14:39:38Z","abstract_excerpt":"We propose TopoOpt, a novel direct-connect fabric for deep neural network (DNN) training workloads. TopoOpt co-optimizes the distributed training process across three dimensions: computation, communication, and network topology. We demonstrate the mutability of AllReduce traffic, and leverage this property to construct efficient network topologies for DNN training jobs. TopoOpt then uses an alternating optimization technique and a group theory-inspired algorithm called TotientPerms to find the best network topology and routing plan, together with a parallelization strategy. We build a fully fu"},"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":"2202.00433","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.NI","submitted_at":"2022-02-01T14:39:38Z","cross_cats_sorted":["cs.DC"],"title_canon_sha256":"707d546977d9972b846f93e5a1025c78d1b6d99825cc76d2947edb66e4c43c6d","abstract_canon_sha256":"b8e7c7944d455e21b6c3725821939a9ad7d3af5d29a8ae31cc8d386d688bc6cc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:02:10.606450Z","signature_b64":"BogNWW4+dZrgClHgDlUYfo9lnkdBZclJdsdx55q0Mhn6r2z+jT6S+9PN5tzY0AAZgGVaUxRduwBnR/Rb30e2CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fb7cc8f0e7d574c50316cb64245e765e9eb550c2bf32826d4fb60dae529da195","last_reissued_at":"2026-07-05T05:02:10.606032Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:02:10.606032Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TopoOpt: Co-optimizing Network Topology and Parallelization Strategy for Distributed Training Jobs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.NI","authors_text":"Anthony Kewitsch, Dheevatsa Mudigere, Manya Ghobadi, Moein Khazraee, Weiyang Wang, Ying Zhang, Zhihao Jia, Zhizhen Zhong","submitted_at":"2022-02-01T14:39:38Z","abstract_excerpt":"We propose TopoOpt, a novel direct-connect fabric for deep neural network (DNN) training workloads. TopoOpt co-optimizes the distributed training process across three dimensions: computation, communication, and network topology. We demonstrate the mutability of AllReduce traffic, and leverage this property to construct efficient network topologies for DNN training jobs. TopoOpt then uses an alternating optimization technique and a group theory-inspired algorithm called TotientPerms to find the best network topology and routing plan, together with a parallelization strategy. We build a fully fu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.00433","kind":"arxiv","version":3},"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/2202.00433/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":"2202.00433","created_at":"2026-07-05T05:02:10.606085+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.00433v3","created_at":"2026-07-05T05:02:10.606085+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.00433","created_at":"2026-07-05T05:02:10.606085+00:00"},{"alias_kind":"pith_short_12","alias_value":"7N6MR4HH2V2M","created_at":"2026-07-05T05:02:10.606085+00:00"},{"alias_kind":"pith_short_16","alias_value":"7N6MR4HH2V2MKAYW","created_at":"2026-07-05T05:02:10.606085+00:00"},{"alias_kind":"pith_short_8","alias_value":"7N6MR4HH","created_at":"2026-07-05T05:02:10.606085+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25954","citing_title":"Step-TP: A Grounded, Step-Level Dataset with Chain-of-Thought Reasoning for LLM-Guided Tensor Program Optimization","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14690","citing_title":"Switching Efficiency: A Novel Framework for Dissecting AI Data Center Network Efficiency","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2","json":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2.json","graph_json":"https://pith.science/api/pith-number/7N6MR4HH2V2MKAYWZNSCIXTWL2/graph.json","events_json":"https://pith.science/api/pith-number/7N6MR4HH2V2MKAYWZNSCIXTWL2/events.json","paper":"https://pith.science/paper/7N6MR4HH"},"agent_actions":{"view_html":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2","download_json":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2.json","view_paper":"https://pith.science/paper/7N6MR4HH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.00433&json=true","fetch_graph":"https://pith.science/api/pith-number/7N6MR4HH2V2MKAYWZNSCIXTWL2/graph.json","fetch_events":"https://pith.science/api/pith-number/7N6MR4HH2V2MKAYWZNSCIXTWL2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2/action/storage_attestation","attest_author":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2/action/author_attestation","sign_citation":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2/action/citation_signature","submit_replication":"https://pith.science/pith/7N6MR4HH2V2MKAYWZNSCIXTWL2/action/replication_record"}},"created_at":"2026-07-05T05:02:10.606085+00:00","updated_at":"2026-07-05T05:02:10.606085+00:00"}