{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DARY36CQKDWILECTIUWVUQ7BFU","short_pith_number":"pith:DARY36CQ","schema_version":"1.0","canonical_sha256":"18238df85050ec859053452d5a43e12d3919806067b3c249693b3823407029f0","source":{"kind":"arxiv","id":"2508.20258","version":1},"attestation_state":"computed","paper":{"title":"SwizzlePerf: Hardware-Aware LLMs for GPU Kernel Performance Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Arya Tschand, Ganesh Dasika, Jeffrey Ma, Keith Lowery, Kesavan Ramakrishnan, Muhammad Awad, Ryan Swann, Vijay Janapa Reddi","submitted_at":"2025-08-27T20:30:43Z","abstract_excerpt":"Large language models (LLMs) have shown progress in GPU kernel performance engineering using inefficient search-based methods that optimize around runtime. Any existing approach lacks a key characteristic that human performance engineers rely on for near-optimal utilization -- hardware-awareness. By leveraging the workload's specific memory access patterns, architecture specifications, filtered profiling logs, and reflections on historical performance, we can make software-level optimizations that are tailored to the underlying hardware. SwizzlePerf automatically generates spatial optimization"},"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":"2508.20258","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2025-08-27T20:30:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"236dacb25714536c1324358bc0873cbf7f9454f340bef4f7b4c54a7d6719cdee","abstract_canon_sha256":"50357c9cdb7d64170950eb326df8f21bc9fb3fd01710ccc0b3de134010dbfc8b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:00:33.104012Z","signature_b64":"VIn9sjH+oE9JZIXNitTh0lKqeoEzCDWV3R/DPDoUmJ0cNsZQe6935hl3LyJfatio8Bo327JE3jxLHcY1ljBFCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"18238df85050ec859053452d5a43e12d3919806067b3c249693b3823407029f0","last_reissued_at":"2026-07-05T12:00:33.103466Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:00:33.103466Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SwizzlePerf: Hardware-Aware LLMs for GPU Kernel Performance Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.DC","authors_text":"Arya Tschand, Ganesh Dasika, Jeffrey Ma, Keith Lowery, Kesavan Ramakrishnan, Muhammad Awad, Ryan Swann, Vijay Janapa Reddi","submitted_at":"2025-08-27T20:30:43Z","abstract_excerpt":"Large language models (LLMs) have shown progress in GPU kernel performance engineering using inefficient search-based methods that optimize around runtime. Any existing approach lacks a key characteristic that human performance engineers rely on for near-optimal utilization -- hardware-awareness. By leveraging the workload's specific memory access patterns, architecture specifications, filtered profiling logs, and reflections on historical performance, we can make software-level optimizations that are tailored to the underlying hardware. SwizzlePerf automatically generates spatial optimization"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.20258","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/2508.20258/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":"2508.20258","created_at":"2026-07-05T12:00:33.103529+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.20258v1","created_at":"2026-07-05T12:00:33.103529+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.20258","created_at":"2026-07-05T12:00:33.103529+00:00"},{"alias_kind":"pith_short_12","alias_value":"DARY36CQKDWI","created_at":"2026-07-05T12:00:33.103529+00:00"},{"alias_kind":"pith_short_16","alias_value":"DARY36CQKDWILECT","created_at":"2026-07-05T12:00:33.103529+00:00"},{"alias_kind":"pith_short_8","alias_value":"DARY36CQ","created_at":"2026-07-05T12:00:33.103529+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17518","citing_title":"SpecGen: Accelerating Agentic Kernel Optimization with Speculative Generation","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30359","citing_title":"Kernel Foundry: A Diagnosis-driven Evolutionary Kernel Optimizer with Multi-Experts","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20032","citing_title":"LEO: Tracing GPU Stall Root Causes via Cross-Vendor Backward Slicing","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU","json":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU.json","graph_json":"https://pith.science/api/pith-number/DARY36CQKDWILECTIUWVUQ7BFU/graph.json","events_json":"https://pith.science/api/pith-number/DARY36CQKDWILECTIUWVUQ7BFU/events.json","paper":"https://pith.science/paper/DARY36CQ"},"agent_actions":{"view_html":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU","download_json":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU.json","view_paper":"https://pith.science/paper/DARY36CQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.20258&json=true","fetch_graph":"https://pith.science/api/pith-number/DARY36CQKDWILECTIUWVUQ7BFU/graph.json","fetch_events":"https://pith.science/api/pith-number/DARY36CQKDWILECTIUWVUQ7BFU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU/action/storage_attestation","attest_author":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU/action/author_attestation","sign_citation":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU/action/citation_signature","submit_replication":"https://pith.science/pith/DARY36CQKDWILECTIUWVUQ7BFU/action/replication_record"}},"created_at":"2026-07-05T12:00:33.103529+00:00","updated_at":"2026-07-05T12:00:33.103529+00:00"}