{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VHHD5NGW66Q4EWTND5RFUML5H5","short_pith_number":"pith:VHHD5NGW","schema_version":"1.0","canonical_sha256":"a9ce3eb4d6f7a1c25a6d1f625a317d3f7f43a8e7cba97ecdd45c3fb1df40ca7b","source":{"kind":"arxiv","id":"2512.09472","version":2},"attestation_state":"computed","paper":{"title":"WarmServe: Enabling One-for-Many GPU Prewarming for Multi-LLM Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Chen Sun, Chiheng Lou, Pengcheng Wang, Rui Kang, Sheng Qi, Xin Jin, Xuanzhe Liu, Yong Zhang","submitted_at":"2025-12-10T09:47:40Z","abstract_excerpt":"Deploying multiple models within shared GPU clusters is a key strategy to improve resource efficiency in large language model (LLM) serving. Existing multi-LLM serving systems improve GPU utilization at the cost of degraded inference performance, particularly time-to-first-token (TTFT). We attribute this degradation to the lack of awareness regarding future workload characteristics. In contrast, recent analyses have shown the strong periodicity and long-term predictability of real-world LLM serving workloads. In this paper, we propose one-for-many GPU prewarming, which proactively loads parame"},"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":"2512.09472","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-12-10T09:47:40Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"cebcc4b2a0041fc529b1603a048ba154dab91f02d181e5141f076f1e07808df0","abstract_canon_sha256":"639b55ad7cf2dd979fb1cdefc16ba87540d07234d0dfd4b7f5b19beaa49d9fb4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:03:52.041985Z","signature_b64":"Nff9zWA17jDDX/QYSIUSk4WEyFlufxFZ5a6SjcitBgoGMm7Hg4tv82xeBIGN5z+v0LgmEeuY6oL82OOYs8VMCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a9ce3eb4d6f7a1c25a6d1f625a317d3f7f43a8e7cba97ecdd45c3fb1df40ca7b","last_reissued_at":"2026-05-22T01:03:52.041121Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:03:52.041121Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"WarmServe: Enabling One-for-Many GPU Prewarming for Multi-LLM Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Chen Sun, Chiheng Lou, Pengcheng Wang, Rui Kang, Sheng Qi, Xin Jin, Xuanzhe Liu, Yong Zhang","submitted_at":"2025-12-10T09:47:40Z","abstract_excerpt":"Deploying multiple models within shared GPU clusters is a key strategy to improve resource efficiency in large language model (LLM) serving. Existing multi-LLM serving systems improve GPU utilization at the cost of degraded inference performance, particularly time-to-first-token (TTFT). We attribute this degradation to the lack of awareness regarding future workload characteristics. In contrast, recent analyses have shown the strong periodicity and long-term predictability of real-world LLM serving workloads. In this paper, we propose one-for-many GPU prewarming, which proactively loads parame"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.09472","kind":"arxiv","version":2},"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/2512.09472/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":"2512.09472","created_at":"2026-05-22T01:03:52.041241+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.09472v2","created_at":"2026-05-22T01:03:52.041241+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.09472","created_at":"2026-05-22T01:03:52.041241+00:00"},{"alias_kind":"pith_short_12","alias_value":"VHHD5NGW66Q4","created_at":"2026-05-22T01:03:52.041241+00:00"},{"alias_kind":"pith_short_16","alias_value":"VHHD5NGW66Q4EWTN","created_at":"2026-05-22T01:03:52.041241+00:00"},{"alias_kind":"pith_short_8","alias_value":"VHHD5NGW","created_at":"2026-05-22T01:03:52.041241+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2603.21354","citing_title":"The Workload-Router-Pool Architecture for LLM Inference Optimization: A Vision Paper from the vLLM Semantic Router Project","ref_index":94,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06664","citing_title":"Foundry: Template-Based CUDA Graph Context Materialization for Fast LLM Serving Cold Start","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5","json":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5.json","graph_json":"https://pith.science/api/pith-number/VHHD5NGW66Q4EWTND5RFUML5H5/graph.json","events_json":"https://pith.science/api/pith-number/VHHD5NGW66Q4EWTND5RFUML5H5/events.json","paper":"https://pith.science/paper/VHHD5NGW"},"agent_actions":{"view_html":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5","download_json":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5.json","view_paper":"https://pith.science/paper/VHHD5NGW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.09472&json=true","fetch_graph":"https://pith.science/api/pith-number/VHHD5NGW66Q4EWTND5RFUML5H5/graph.json","fetch_events":"https://pith.science/api/pith-number/VHHD5NGW66Q4EWTND5RFUML5H5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5/action/storage_attestation","attest_author":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5/action/author_attestation","sign_citation":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5/action/citation_signature","submit_replication":"https://pith.science/pith/VHHD5NGW66Q4EWTND5RFUML5H5/action/replication_record"}},"created_at":"2026-05-22T01:03:52.041241+00:00","updated_at":"2026-05-22T01:03:52.041241+00:00"}