{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NJC3P2HVMBE45EDJUWCTJQEFTD","short_pith_number":"pith:NJC3P2HV","schema_version":"1.0","canonical_sha256":"6a45b7e8f56049ce9069a58534c08598e7767da3428228736c58feaf96bc3f01","source":{"kind":"arxiv","id":"2605.28302","version":1},"attestation_state":"computed","paper":{"title":"How Far Can Disaggregation Go? A Design-Space Exploration of Attention-FFN Disaggregation for Efficient MoE LLM Serving","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.DC"],"primary_cat":"cs.LG","authors_text":"Abhimanyu Rajeshkumar Bambhaniya, Amir Yazdanbakhsh, Hanjiang Wu, Madhu Kumar, Midhilesh Elavazhagan, Sarbartha Banerjee, Souvik Kundu, Sudarshan Srinivasan, Suvinay Subramanian, Tuhin Khare, Tushar Krishna, William Won","submitted_at":"2026-05-27T10:55:57Z","abstract_excerpt":"Modern large language model (LLM) inference has progressively disaggregated to keep pace with growing model sizes and tight TTFT and TPOT service-level objectives: from chunked-prefill aggregation, to prefill-decode (P/D) disaggregation, and most recently to operator-level Attention-FFN Disaggregation (AFD). This trend is especially important for mixture-of-experts (MoE) models, where memory-bound attention, compute-intensive expert FFNs, and MoE dispatch/combine communication create distinct resource demands. AFD further exposes this heterogeneity by placing attention and MoE-FFN execution on"},"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":"2605.28302","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T10:55:57Z","cross_cats_sorted":["cs.AI","cs.DC"],"title_canon_sha256":"1a145082f61032dbe15bcd6f9e983b76cecd7a23cb828d38701f658efaf61c0b","abstract_canon_sha256":"043b7ae2828c610efb865e1a88d3521a25648cfdb0e65aaeabcf8540d4c93d12"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T01:05:05.774387Z","signature_b64":"CSVYLzU8J5zi8WxhL+kVMC3ADr0rKD+dEXEOsszwHUORnHtWatH5lPK0zRnEjlNRNHczYIWNPe3WpNL0tU+uCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6a45b7e8f56049ce9069a58534c08598e7767da3428228736c58feaf96bc3f01","last_reissued_at":"2026-05-28T01:05:05.773951Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T01:05:05.773951Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How Far Can Disaggregation Go? A Design-Space Exploration of Attention-FFN Disaggregation for Efficient MoE LLM Serving","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.DC"],"primary_cat":"cs.LG","authors_text":"Abhimanyu Rajeshkumar Bambhaniya, Amir Yazdanbakhsh, Hanjiang Wu, Madhu Kumar, Midhilesh Elavazhagan, Sarbartha Banerjee, Souvik Kundu, Sudarshan Srinivasan, Suvinay Subramanian, Tuhin Khare, Tushar Krishna, William Won","submitted_at":"2026-05-27T10:55:57Z","abstract_excerpt":"Modern large language model (LLM) inference has progressively disaggregated to keep pace with growing model sizes and tight TTFT and TPOT service-level objectives: from chunked-prefill aggregation, to prefill-decode (P/D) disaggregation, and most recently to operator-level Attention-FFN Disaggregation (AFD). This trend is especially important for mixture-of-experts (MoE) models, where memory-bound attention, compute-intensive expert FFNs, and MoE dispatch/combine communication create distinct resource demands. AFD further exposes this heterogeneity by placing attention and MoE-FFN execution on"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28302","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/2605.28302/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":"2605.28302","created_at":"2026-05-28T01:05:05.774019+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.28302v1","created_at":"2026-05-28T01:05:05.774019+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28302","created_at":"2026-05-28T01:05:05.774019+00:00"},{"alias_kind":"pith_short_12","alias_value":"NJC3P2HVMBE4","created_at":"2026-05-28T01:05:05.774019+00:00"},{"alias_kind":"pith_short_16","alias_value":"NJC3P2HVMBE45EDJ","created_at":"2026-05-28T01:05:05.774019+00:00"},{"alias_kind":"pith_short_8","alias_value":"NJC3P2HV","created_at":"2026-05-28T01:05:05.774019+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/NJC3P2HVMBE45EDJUWCTJQEFTD","json":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD.json","graph_json":"https://pith.science/api/pith-number/NJC3P2HVMBE45EDJUWCTJQEFTD/graph.json","events_json":"https://pith.science/api/pith-number/NJC3P2HVMBE45EDJUWCTJQEFTD/events.json","paper":"https://pith.science/paper/NJC3P2HV"},"agent_actions":{"view_html":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD","download_json":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD.json","view_paper":"https://pith.science/paper/NJC3P2HV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.28302&json=true","fetch_graph":"https://pith.science/api/pith-number/NJC3P2HVMBE45EDJUWCTJQEFTD/graph.json","fetch_events":"https://pith.science/api/pith-number/NJC3P2HVMBE45EDJUWCTJQEFTD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD/action/storage_attestation","attest_author":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD/action/author_attestation","sign_citation":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD/action/citation_signature","submit_replication":"https://pith.science/pith/NJC3P2HVMBE45EDJUWCTJQEFTD/action/replication_record"}},"created_at":"2026-05-28T01:05:05.774019+00:00","updated_at":"2026-05-28T01:05:05.774019+00:00"}