{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UYC5UHOXXBZXRB7VC5CQG6DLX3","short_pith_number":"pith:UYC5UHOX","schema_version":"1.0","canonical_sha256":"a605da1dd7b8737887f5174503786bbef4e6b77afcbe80cfcf1955b6be1e2c5e","source":{"kind":"arxiv","id":"2607.17074","version":1},"attestation_state":"computed","paper":{"title":"ThAME: 3D Memory-Enabled Heterogeneous Accelerator for LLM Mixture of Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DC"],"primary_cat":"cs.AR","authors_text":"Janardhan Rao Doppa, Partha Pratim Pande, Pramit Kumar Pal, Pratyush Dhingra","submitted_at":"2026-07-19T04:45:01Z","abstract_excerpt":"Mixture of Experts (MoE) architectures have emerged as a dominant paradigm for scaling Large Language Models (LLMs). However, MoE inference on conventional hardware is constrained by three fundamental bottlenecks. These encompass the massive memory bandwidth required to fetch non-contiguous expert weights, the non-deterministic scatter-gather traffic generated by input-dependent token routing, and the tail-latency dependency imposed by synchronous expert output aggregation. To address these challenges, we propose ThAME, a three-dimensional (3D) heterogeneous multi-chiplet architecture for MoE "},"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":"2607.17074","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2026-07-19T04:45:01Z","cross_cats_sorted":["cs.AI","cs.DC"],"title_canon_sha256":"b6ddd3e17c6c242f5e3cac894fac612a88deb86b0ccc7b5b0c7b552e5dcd9c22","abstract_canon_sha256":"62e099de86d50b2a65257c9ed7f0a5398c82e3be46b2645b53de1da2e6445a1d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T01:21:13.514106Z","signature_b64":"IWB+5GHxjPj6adnFIZ+kUD465nX0McTt3V8WRebF7x7dgyTBq230hCSnDDZD90MrKgGN9gumcyZ4d4YXkeQMCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a605da1dd7b8737887f5174503786bbef4e6b77afcbe80cfcf1955b6be1e2c5e","last_reissued_at":"2026-07-21T01:21:13.513254Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T01:21:13.513254Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ThAME: 3D Memory-Enabled Heterogeneous Accelerator for LLM Mixture of Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DC"],"primary_cat":"cs.AR","authors_text":"Janardhan Rao Doppa, Partha Pratim Pande, Pramit Kumar Pal, Pratyush Dhingra","submitted_at":"2026-07-19T04:45:01Z","abstract_excerpt":"Mixture of Experts (MoE) architectures have emerged as a dominant paradigm for scaling Large Language Models (LLMs). However, MoE inference on conventional hardware is constrained by three fundamental bottlenecks. These encompass the massive memory bandwidth required to fetch non-contiguous expert weights, the non-deterministic scatter-gather traffic generated by input-dependent token routing, and the tail-latency dependency imposed by synchronous expert output aggregation. To address these challenges, we propose ThAME, a three-dimensional (3D) heterogeneous multi-chiplet architecture for MoE "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.17074","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/2607.17074/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":"2607.17074","created_at":"2026-07-21T01:21:13.513699+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.17074v1","created_at":"2026-07-21T01:21:13.513699+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.17074","created_at":"2026-07-21T01:21:13.513699+00:00"},{"alias_kind":"pith_short_12","alias_value":"UYC5UHOXXBZX","created_at":"2026-07-21T01:21:13.513699+00:00"},{"alias_kind":"pith_short_16","alias_value":"UYC5UHOXXBZXRB7V","created_at":"2026-07-21T01:21:13.513699+00:00"},{"alias_kind":"pith_short_8","alias_value":"UYC5UHOX","created_at":"2026-07-21T01:21:13.513699+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/UYC5UHOXXBZXRB7VC5CQG6DLX3","json":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3.json","graph_json":"https://pith.science/api/pith-number/UYC5UHOXXBZXRB7VC5CQG6DLX3/graph.json","events_json":"https://pith.science/api/pith-number/UYC5UHOXXBZXRB7VC5CQG6DLX3/events.json","paper":"https://pith.science/paper/UYC5UHOX"},"agent_actions":{"view_html":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3","download_json":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3.json","view_paper":"https://pith.science/paper/UYC5UHOX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.17074&json=true","fetch_graph":"https://pith.science/api/pith-number/UYC5UHOXXBZXRB7VC5CQG6DLX3/graph.json","fetch_events":"https://pith.science/api/pith-number/UYC5UHOXXBZXRB7VC5CQG6DLX3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3/action/storage_attestation","attest_author":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3/action/author_attestation","sign_citation":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3/action/citation_signature","submit_replication":"https://pith.science/pith/UYC5UHOXXBZXRB7VC5CQG6DLX3/action/replication_record"}},"created_at":"2026-07-21T01:21:13.513699+00:00","updated_at":"2026-07-21T01:21:13.513699+00:00"}