{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GIOKATMP3M6GL7CSRUDVBU6GNC","short_pith_number":"pith:GIOKATMP","schema_version":"1.0","canonical_sha256":"321ca04d8fdb3c65fc528d0750d3c668a8671e04f813d3f79cfaac8856104d0f","source":{"kind":"arxiv","id":"2602.02987","version":2},"attestation_state":"computed","paper":{"title":"Large-Scale LLM Inference with Heterogeneous Workloads: Prefill-Decode Contention and Asymptotically Optimal Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.DC","authors_text":"Jiheng Zhang, Ruihan Lin, Zean Han, Zezhen Ding","submitted_at":"2026-02-03T01:47:37Z","abstract_excerpt":"Large Language Models (LLMs) are rapidly becoming critical infrastructure for enterprise applications, driving unprecedented demand for GPU-based inference services. A key operational challenge arises from the two-phase nature of LLM inference: a compute-intensive \\emph{prefill} phase that processes user input, followed by a memory-bound \\emph{decode} phase that generates output tokens. When these phases share GPU resources, prefill tasks throttle the processing speed of concurrent decodes, creating state-dependent contention. This contention is further complicated by workload heterogeneity, a"},"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":"2602.02987","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2026-02-03T01:47:37Z","cross_cats_sorted":["math.OC"],"title_canon_sha256":"5365ecd1bdc179db4d45051d1a4f22a92728778a969b4c9064b38bce1dbdae38","abstract_canon_sha256":"8c4bd1952e705e13c95e6a83d3a2593979622fa6a252883d8808c216cc66ea76"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:14:34.250659Z","signature_b64":"GA1XfuCP1VDHgaBE9aCvr9fXXUPxXps1KBj45mye4VIg7yW9GnnNAOEl1hLeoFj9Y862dgnM8VHq7/v66Dw6AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"321ca04d8fdb3c65fc528d0750d3c668a8671e04f813d3f79cfaac8856104d0f","last_reissued_at":"2026-06-05T01:14:34.249796Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:14:34.249796Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large-Scale LLM Inference with Heterogeneous Workloads: Prefill-Decode Contention and Asymptotically Optimal Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.DC","authors_text":"Jiheng Zhang, Ruihan Lin, Zean Han, Zezhen Ding","submitted_at":"2026-02-03T01:47:37Z","abstract_excerpt":"Large Language Models (LLMs) are rapidly becoming critical infrastructure for enterprise applications, driving unprecedented demand for GPU-based inference services. A key operational challenge arises from the two-phase nature of LLM inference: a compute-intensive \\emph{prefill} phase that processes user input, followed by a memory-bound \\emph{decode} phase that generates output tokens. When these phases share GPU resources, prefill tasks throttle the processing speed of concurrent decodes, creating state-dependent contention. This contention is further complicated by workload heterogeneity, a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.02987","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/2602.02987/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":"2602.02987","created_at":"2026-06-05T01:14:34.249920+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.02987v2","created_at":"2026-06-05T01:14:34.249920+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.02987","created_at":"2026-06-05T01:14:34.249920+00:00"},{"alias_kind":"pith_short_12","alias_value":"GIOKATMP3M6G","created_at":"2026-06-05T01:14:34.249920+00:00"},{"alias_kind":"pith_short_16","alias_value":"GIOKATMP3M6GL7CS","created_at":"2026-06-05T01:14:34.249920+00:00"},{"alias_kind":"pith_short_8","alias_value":"GIOKATMP","created_at":"2026-06-05T01:14:34.249920+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/GIOKATMP3M6GL7CSRUDVBU6GNC","json":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC.json","graph_json":"https://pith.science/api/pith-number/GIOKATMP3M6GL7CSRUDVBU6GNC/graph.json","events_json":"https://pith.science/api/pith-number/GIOKATMP3M6GL7CSRUDVBU6GNC/events.json","paper":"https://pith.science/paper/GIOKATMP"},"agent_actions":{"view_html":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC","download_json":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC.json","view_paper":"https://pith.science/paper/GIOKATMP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.02987&json=true","fetch_graph":"https://pith.science/api/pith-number/GIOKATMP3M6GL7CSRUDVBU6GNC/graph.json","fetch_events":"https://pith.science/api/pith-number/GIOKATMP3M6GL7CSRUDVBU6GNC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC/action/storage_attestation","attest_author":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC/action/author_attestation","sign_citation":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC/action/citation_signature","submit_replication":"https://pith.science/pith/GIOKATMP3M6GL7CSRUDVBU6GNC/action/replication_record"}},"created_at":"2026-06-05T01:14:34.249920+00:00","updated_at":"2026-06-05T01:14:34.249920+00:00"}