{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UKVSG4PBZCCPHBLOQXC4U4EU5Q","short_pith_number":"pith:UKVSG4PB","schema_version":"1.0","canonical_sha256":"a2ab2371e1c884f3856e85c5ca7094ec1497f526ba56db1ae224b5e984aade7d","source":{"kind":"arxiv","id":"2606.11357","version":1},"attestation_state":"computed","paper":{"title":"TileFuse: A Fused Mixed-Precision Kernel Library for Efficient Quantized LLM Inference on AMD NPUs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.AR","cs.PF"],"primary_cat":"cs.DC","authors_text":"Deming Chen, Feiyang Liu, Gregory Hyegang Jun, Wesley Pang","submitted_at":"2026-06-09T18:33:14Z","abstract_excerpt":"With the growing demand for on-device LLM inference, edge SoCs increasingly integrate NPUs to improve performance and energy efficiency under tight power and thermal budgets. However, practical LLM deployment on current client NPUs remains difficult: widely used quantization formats such as AWQ do not map cleanly onto many existing NPU software stacks, which are often proprietary and expose limited low-level control. In this work, we present \\textit{TileFuse}, a close-to-metal mixed-precision kernel library for AMD XDNA2 NPUs that targets transformer linear layers in quantized LLM inference. T"},"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":"2606.11357","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-06-09T18:33:14Z","cross_cats_sorted":["cs.AI","cs.AR","cs.PF"],"title_canon_sha256":"774bd3a2d255a97c8f7d41c440b7bda32e08333717831524a7a16eed939d6729","abstract_canon_sha256":"67911681fa078e51da5d630a26cd7ab23dd096a7fab1fc79319a999899b2a153"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-11T00:08:21.454138Z","signature_b64":"mXrWP0PsKpaI+K4DWngzskV5a+Z6GQSks03qXDfH3zKEorS//ng+nALZfqa/5/n27WXNADjEEeP2uv08odhRBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a2ab2371e1c884f3856e85c5ca7094ec1497f526ba56db1ae224b5e984aade7d","last_reissued_at":"2026-06-11T00:08:21.453263Z","signature_status":"signed_v1","first_computed_at":"2026-06-11T00:08:21.453263Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TileFuse: A Fused Mixed-Precision Kernel Library for Efficient Quantized LLM Inference on AMD NPUs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.AR","cs.PF"],"primary_cat":"cs.DC","authors_text":"Deming Chen, Feiyang Liu, Gregory Hyegang Jun, Wesley Pang","submitted_at":"2026-06-09T18:33:14Z","abstract_excerpt":"With the growing demand for on-device LLM inference, edge SoCs increasingly integrate NPUs to improve performance and energy efficiency under tight power and thermal budgets. However, practical LLM deployment on current client NPUs remains difficult: widely used quantization formats such as AWQ do not map cleanly onto many existing NPU software stacks, which are often proprietary and expose limited low-level control. In this work, we present \\textit{TileFuse}, a close-to-metal mixed-precision kernel library for AMD XDNA2 NPUs that targets transformer linear layers in quantized LLM inference. T"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.11357","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/2606.11357/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":"2606.11357","created_at":"2026-06-11T00:08:21.453419+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.11357v1","created_at":"2026-06-11T00:08:21.453419+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.11357","created_at":"2026-06-11T00:08:21.453419+00:00"},{"alias_kind":"pith_short_12","alias_value":"UKVSG4PBZCCP","created_at":"2026-06-11T00:08:21.453419+00:00"},{"alias_kind":"pith_short_16","alias_value":"UKVSG4PBZCCPHBLO","created_at":"2026-06-11T00:08:21.453419+00:00"},{"alias_kind":"pith_short_8","alias_value":"UKVSG4PB","created_at":"2026-06-11T00:08:21.453419+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/UKVSG4PBZCCPHBLOQXC4U4EU5Q","json":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q.json","graph_json":"https://pith.science/api/pith-number/UKVSG4PBZCCPHBLOQXC4U4EU5Q/graph.json","events_json":"https://pith.science/api/pith-number/UKVSG4PBZCCPHBLOQXC4U4EU5Q/events.json","paper":"https://pith.science/paper/UKVSG4PB"},"agent_actions":{"view_html":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q","download_json":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q.json","view_paper":"https://pith.science/paper/UKVSG4PB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.11357&json=true","fetch_graph":"https://pith.science/api/pith-number/UKVSG4PBZCCPHBLOQXC4U4EU5Q/graph.json","fetch_events":"https://pith.science/api/pith-number/UKVSG4PBZCCPHBLOQXC4U4EU5Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q/action/storage_attestation","attest_author":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q/action/author_attestation","sign_citation":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q/action/citation_signature","submit_replication":"https://pith.science/pith/UKVSG4PBZCCPHBLOQXC4U4EU5Q/action/replication_record"}},"created_at":"2026-06-11T00:08:21.453419+00:00","updated_at":"2026-06-11T00:08:21.453419+00:00"}