{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IUWLQCSGFUWDECUSQPTIYEQ6ZQ","short_pith_number":"pith:IUWLQCSG","schema_version":"1.0","canonical_sha256":"452cb80a462d2c320a9283e68c121ecc1c175502e5107e0b812b9dc7ddd880de","source":{"kind":"arxiv","id":"2511.04791","version":2},"attestation_state":"computed","paper":{"title":"DuetServe: Harmonizing Prefill and Decode for LLM Serving via Adaptive GPU Multiplexing","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chaoyi Jiang, Daniel Wong, Hossein Entezari Zarch, Lei Gao, Mark Hill, Murali Annavaram","submitted_at":"2025-11-06T20:18:34Z","abstract_excerpt":"Modern LLM serving systems must sustain high throughput while meeting strict latency SLOs across two distinct inference phases: compute-intensive prefill and memory-bound decode phases. Existing approaches either (1) aggregate both phases on shared GPUs, leading to interference between prefill and decode phases, which degrades Time-Between-Tokens (TBT); or (2) disaggregate the two phases across GPUs, improving latency but wasting resources through duplicated models and KV cache transfers. We present DuetServe, a unified LLM serving framework that achieves disaggregation-level isolation within "},"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":"2511.04791","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-06T20:18:34Z","cross_cats_sorted":[],"title_canon_sha256":"08a017afb2a6c36ac126bb9975114dc2b9d20b4cd0f7795d48dcf78ffa08a6ca","abstract_canon_sha256":"64ccf35c7aba444f8cbfad5c68d0d28474036eceec2be88b3b13303bd2170317"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:37.951702Z","signature_b64":"RDRkzNrIAcsw1KW7fHShBhh+3e1IE9W0u4R/tUisyDogXYOYp2PZkjbMGqdQ1wsRjm3lVvWrwmqruSj0ngRfBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"452cb80a462d2c320a9283e68c121ecc1c175502e5107e0b812b9dc7ddd880de","last_reissued_at":"2026-06-02T01:03:37.951027Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:37.951027Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DuetServe: Harmonizing Prefill and Decode for LLM Serving via Adaptive GPU Multiplexing","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chaoyi Jiang, Daniel Wong, Hossein Entezari Zarch, Lei Gao, Mark Hill, Murali Annavaram","submitted_at":"2025-11-06T20:18:34Z","abstract_excerpt":"Modern LLM serving systems must sustain high throughput while meeting strict latency SLOs across two distinct inference phases: compute-intensive prefill and memory-bound decode phases. Existing approaches either (1) aggregate both phases on shared GPUs, leading to interference between prefill and decode phases, which degrades Time-Between-Tokens (TBT); or (2) disaggregate the two phases across GPUs, improving latency but wasting resources through duplicated models and KV cache transfers. We present DuetServe, a unified LLM serving framework that achieves disaggregation-level isolation within "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.04791","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/2511.04791/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":"2511.04791","created_at":"2026-06-02T01:03:37.951100+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.04791v2","created_at":"2026-06-02T01:03:37.951100+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.04791","created_at":"2026-06-02T01:03:37.951100+00:00"},{"alias_kind":"pith_short_12","alias_value":"IUWLQCSGFUWD","created_at":"2026-06-02T01:03:37.951100+00:00"},{"alias_kind":"pith_short_16","alias_value":"IUWLQCSGFUWDECUS","created_at":"2026-06-02T01:03:37.951100+00:00"},{"alias_kind":"pith_short_8","alias_value":"IUWLQCSG","created_at":"2026-06-02T01:03:37.951100+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/IUWLQCSGFUWDECUSQPTIYEQ6ZQ","json":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ.json","graph_json":"https://pith.science/api/pith-number/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/graph.json","events_json":"https://pith.science/api/pith-number/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/events.json","paper":"https://pith.science/paper/IUWLQCSG"},"agent_actions":{"view_html":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ","download_json":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ.json","view_paper":"https://pith.science/paper/IUWLQCSG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.04791&json=true","fetch_graph":"https://pith.science/api/pith-number/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/graph.json","fetch_events":"https://pith.science/api/pith-number/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/action/storage_attestation","attest_author":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/action/author_attestation","sign_citation":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/action/citation_signature","submit_replication":"https://pith.science/pith/IUWLQCSGFUWDECUSQPTIYEQ6ZQ/action/replication_record"}},"created_at":"2026-06-02T01:03:37.951100+00:00","updated_at":"2026-06-02T01:03:37.951100+00:00"}