{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YRUNQITD2RKQXO2YH7SDTYYAQ3","short_pith_number":"pith:YRUNQITD","schema_version":"1.0","canonical_sha256":"c468d82263d4550bbb583fe439e30086cc833a1441952277c5cded0553e79d5c","source":{"kind":"arxiv","id":"2407.14645","version":1},"attestation_state":"computed","paper":{"title":"Performance Modeling and Workload Analysis of Distributed Large Language Model Training and Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC","cs.LG"],"primary_cat":"cs.AR","authors_text":"Ali BanaGozar, Arindam Mallik, Joyjit Kundu, Puneet Gupta, Sourav Sengupta, Udari De Alwis, Wenzhe Guo","submitted_at":"2024-07-19T19:49:05Z","abstract_excerpt":"Aligning future system design with the ever-increasing compute needs of large language models (LLMs) is undoubtedly an important problem in today's world. Here, we propose a general performance modeling methodology and workload analysis of distributed LLM training and inference through an analytical framework that accurately considers compute, memory sub-system, network, and various parallelization strategies (model parallel, data parallel, pipeline parallel, and sequence parallel). We validate our performance predictions with published data from literature and relevant industry vendors (e.g.,"},"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":"2407.14645","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2024-07-19T19:49:05Z","cross_cats_sorted":["cs.DC","cs.LG"],"title_canon_sha256":"a90743bc5c74657aa603ed478079b4dda172f9ff8c715ca8070fa70e482b1c20","abstract_canon_sha256":"d072e86f74f00b898881e2b453da8b9295c79b1a27e69c0bd559f9979db46e09"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:46:31.670385Z","signature_b64":"iekUcAhqlsXYJHuZ6BWiwyRpvyVCpil5OwYWAyLLn45lV6ZMRfTctOgg26rwUKvFdQgC7ZLrlMMou5vrn6HLAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c468d82263d4550bbb583fe439e30086cc833a1441952277c5cded0553e79d5c","last_reissued_at":"2026-07-05T08:46:31.669922Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:46:31.669922Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Performance Modeling and Workload Analysis of Distributed Large Language Model Training and Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC","cs.LG"],"primary_cat":"cs.AR","authors_text":"Ali BanaGozar, Arindam Mallik, Joyjit Kundu, Puneet Gupta, Sourav Sengupta, Udari De Alwis, Wenzhe Guo","submitted_at":"2024-07-19T19:49:05Z","abstract_excerpt":"Aligning future system design with the ever-increasing compute needs of large language models (LLMs) is undoubtedly an important problem in today's world. Here, we propose a general performance modeling methodology and workload analysis of distributed LLM training and inference through an analytical framework that accurately considers compute, memory sub-system, network, and various parallelization strategies (model parallel, data parallel, pipeline parallel, and sequence parallel). We validate our performance predictions with published data from literature and relevant industry vendors (e.g.,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.14645","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/2407.14645/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":"2407.14645","created_at":"2026-07-05T08:46:31.669979+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.14645v1","created_at":"2026-07-05T08:46:31.669979+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.14645","created_at":"2026-07-05T08:46:31.669979+00:00"},{"alias_kind":"pith_short_12","alias_value":"YRUNQITD2RKQ","created_at":"2026-07-05T08:46:31.669979+00:00"},{"alias_kind":"pith_short_16","alias_value":"YRUNQITD2RKQXO2Y","created_at":"2026-07-05T08:46:31.669979+00:00"},{"alias_kind":"pith_short_8","alias_value":"YRUNQITD","created_at":"2026-07-05T08:46:31.669979+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.05876","citing_title":"Think Before You Grid-Search: Floor-First Triage for LLM Serving","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2607.05876","citing_title":"Think Before You Grid-Search: Floor-First Triage for LLM Serving","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2604.12090","citing_title":"Evaluating Cross-Architecture Performance Modeling of Distributed ML Workloads Using StableHLO","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3","json":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3.json","graph_json":"https://pith.science/api/pith-number/YRUNQITD2RKQXO2YH7SDTYYAQ3/graph.json","events_json":"https://pith.science/api/pith-number/YRUNQITD2RKQXO2YH7SDTYYAQ3/events.json","paper":"https://pith.science/paper/YRUNQITD"},"agent_actions":{"view_html":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3","download_json":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3.json","view_paper":"https://pith.science/paper/YRUNQITD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.14645&json=true","fetch_graph":"https://pith.science/api/pith-number/YRUNQITD2RKQXO2YH7SDTYYAQ3/graph.json","fetch_events":"https://pith.science/api/pith-number/YRUNQITD2RKQXO2YH7SDTYYAQ3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3/action/storage_attestation","attest_author":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3/action/author_attestation","sign_citation":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3/action/citation_signature","submit_replication":"https://pith.science/pith/YRUNQITD2RKQXO2YH7SDTYYAQ3/action/replication_record"}},"created_at":"2026-07-05T08:46:31.669979+00:00","updated_at":"2026-07-05T08:46:31.669979+00:00"}