{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7JIM3QZWRHTBXWRS4OOZPU7URB","short_pith_number":"pith:7JIM3QZW","schema_version":"1.0","canonical_sha256":"fa50cdc33689e61bda32e39d97d3f48859ac9a31f2afa50d868818a05b311f4b","source":{"kind":"arxiv","id":"2507.09010","version":1},"attestation_state":"computed","paper":{"title":"Hybrid Systolic Array Accelerator with Optimized Dataflow for Edge Large Language Model Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.AR","authors_text":"Chun-Ting Chen, HanGyeol Mun, Jae-sun Seo, Jian Meng, Mohamed S. Abdelfattah","submitted_at":"2025-07-11T20:27:30Z","abstract_excerpt":"Edge inference for large language models (LLM) offers secure, low-latency, and cost-effective inference solutions. We emphasize that an edge accelerator should achieve high area efficiency and minimize external memory access (EMA) during the memory-bound decode stage, while maintaining high energy efficiency during the compute intensive prefill stage. This paper proposes an edge LLM inference accelerator featuring a hybrid systolic array (HSA) architecture that optimizes inference efficiency in both stages. To further reduce EMA, we adopt MXINT4 weight quantization and propose an optimized dat"},"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":"2507.09010","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2025-07-11T20:27:30Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e22e0fd0ea8615fd4fbc34386e4cdbe68b45256a3c00024f439162ecd077ea7d","abstract_canon_sha256":"77c41040319c9c9d119f54b0d255f12c89d0d07e3d2fee505ebac24512530ec4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:36:02.148520Z","signature_b64":"O+cjF3yV3cGTA8b52a4vB9pNi9SXYgCdX28yGc5LoeF730HP7J5uhHGxE9xGK6M4wxqutCe0vESammIJN4gfAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fa50cdc33689e61bda32e39d97d3f48859ac9a31f2afa50d868818a05b311f4b","last_reissued_at":"2026-07-05T11:36:02.148064Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:36:02.148064Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hybrid Systolic Array Accelerator with Optimized Dataflow for Edge Large Language Model Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.AR","authors_text":"Chun-Ting Chen, HanGyeol Mun, Jae-sun Seo, Jian Meng, Mohamed S. Abdelfattah","submitted_at":"2025-07-11T20:27:30Z","abstract_excerpt":"Edge inference for large language models (LLM) offers secure, low-latency, and cost-effective inference solutions. We emphasize that an edge accelerator should achieve high area efficiency and minimize external memory access (EMA) during the memory-bound decode stage, while maintaining high energy efficiency during the compute intensive prefill stage. This paper proposes an edge LLM inference accelerator featuring a hybrid systolic array (HSA) architecture that optimizes inference efficiency in both stages. To further reduce EMA, we adopt MXINT4 weight quantization and propose an optimized dat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.09010","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/2507.09010/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":"2507.09010","created_at":"2026-07-05T11:36:02.148115+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.09010v1","created_at":"2026-07-05T11:36:02.148115+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.09010","created_at":"2026-07-05T11:36:02.148115+00:00"},{"alias_kind":"pith_short_12","alias_value":"7JIM3QZWRHTB","created_at":"2026-07-05T11:36:02.148115+00:00"},{"alias_kind":"pith_short_16","alias_value":"7JIM3QZWRHTBXWRS","created_at":"2026-07-05T11:36:02.148115+00:00"},{"alias_kind":"pith_short_8","alias_value":"7JIM3QZW","created_at":"2026-07-05T11:36:02.148115+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/7JIM3QZWRHTBXWRS4OOZPU7URB","json":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB.json","graph_json":"https://pith.science/api/pith-number/7JIM3QZWRHTBXWRS4OOZPU7URB/graph.json","events_json":"https://pith.science/api/pith-number/7JIM3QZWRHTBXWRS4OOZPU7URB/events.json","paper":"https://pith.science/paper/7JIM3QZW"},"agent_actions":{"view_html":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB","download_json":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB.json","view_paper":"https://pith.science/paper/7JIM3QZW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.09010&json=true","fetch_graph":"https://pith.science/api/pith-number/7JIM3QZWRHTBXWRS4OOZPU7URB/graph.json","fetch_events":"https://pith.science/api/pith-number/7JIM3QZWRHTBXWRS4OOZPU7URB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB/action/storage_attestation","attest_author":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB/action/author_attestation","sign_citation":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB/action/citation_signature","submit_replication":"https://pith.science/pith/7JIM3QZWRHTBXWRS4OOZPU7URB/action/replication_record"}},"created_at":"2026-07-05T11:36:02.148115+00:00","updated_at":"2026-07-05T11:36:02.148115+00:00"}