{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7PHI5YKCK3JHDMDUKP6VLU655V","short_pith_number":"pith:7PHI5YKC","schema_version":"1.0","canonical_sha256":"fbce8ee14256d271b07453fd55d3dded54c35a4a7d4376335729b9356b82d4eb","source":{"kind":"arxiv","id":"2608.03537","version":1},"attestation_state":"computed","paper":{"title":"ComFuse: Fusing Complex Memory-Intensive Subgraphs with Compute-Intensive Kernels For Modern GPU Architectures","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AR","authors_text":"Di Mu, Gang Wang, Jialin Yang, Mian Huo, Shusong Guo, Tengyuan Jin, Xiaoguang Liu, Yusen Li, Zhenkun Wang","submitted_at":"2026-08-04T12:15:28Z","abstract_excerpt":"Modern deep learning workloads increasingly comprise heterogeneous computation graphs that combine compute-intensive operators with memory-intensive subgraphs. Existing deep learning compilers typically optimize these operator classes separately, creating rigid fusion boundaries that limit cross-operator optimization and on-chip data reuse. We observe that downstream memory-intensive operations can execute concurrently with compute-intensive operators, allowing their execution to be hidden behind computation; however, automatically exploiting this opportunity poses new compilation challenges.\n"},"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":"2608.03537","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2026-08-04T12:15:28Z","cross_cats_sorted":[],"title_canon_sha256":"534423dd2eefc618b221f0d7c32d776ee33e96f132e1169076ff16752d10a2db","abstract_canon_sha256":"bd8d7f314fef2ce1302ca2881934d87f9872c2e1f64ff2cc1a58e1e840f338ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-05T01:36:03.994785Z","signature_b64":"ui/N+VkFXKiIfPx4PqU/dGmVXU2HUsn1wuTFgj709GALYbwF/Slo0KPAURhFX8NPhKyFlZ5Y7Gde/cp+8cIgBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fbce8ee14256d271b07453fd55d3dded54c35a4a7d4376335729b9356b82d4eb","last_reissued_at":"2026-08-05T01:36:03.993378Z","signature_status":"signed_v1","first_computed_at":"2026-08-05T01:36:03.993378Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ComFuse: Fusing Complex Memory-Intensive Subgraphs with Compute-Intensive Kernels For Modern GPU Architectures","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AR","authors_text":"Di Mu, Gang Wang, Jialin Yang, Mian Huo, Shusong Guo, Tengyuan Jin, Xiaoguang Liu, Yusen Li, Zhenkun Wang","submitted_at":"2026-08-04T12:15:28Z","abstract_excerpt":"Modern deep learning workloads increasingly comprise heterogeneous computation graphs that combine compute-intensive operators with memory-intensive subgraphs. Existing deep learning compilers typically optimize these operator classes separately, creating rigid fusion boundaries that limit cross-operator optimization and on-chip data reuse. We observe that downstream memory-intensive operations can execute concurrently with compute-intensive operators, allowing their execution to be hidden behind computation; however, automatically exploiting this opportunity poses new compilation challenges.\n"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.03537","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/2608.03537/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":"2608.03537","created_at":"2026-08-05T01:36:03.993818+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.03537v1","created_at":"2026-08-05T01:36:03.993818+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.03537","created_at":"2026-08-05T01:36:03.993818+00:00"},{"alias_kind":"pith_short_12","alias_value":"7PHI5YKCK3JH","created_at":"2026-08-05T01:36:03.993818+00:00"},{"alias_kind":"pith_short_16","alias_value":"7PHI5YKCK3JHDMDU","created_at":"2026-08-05T01:36:03.993818+00:00"},{"alias_kind":"pith_short_8","alias_value":"7PHI5YKC","created_at":"2026-08-05T01:36:03.993818+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/7PHI5YKCK3JHDMDUKP6VLU655V","json":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V.json","graph_json":"https://pith.science/api/pith-number/7PHI5YKCK3JHDMDUKP6VLU655V/graph.json","events_json":"https://pith.science/api/pith-number/7PHI5YKCK3JHDMDUKP6VLU655V/events.json","paper":"https://pith.science/paper/7PHI5YKC"},"agent_actions":{"view_html":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V","download_json":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V.json","view_paper":"https://pith.science/paper/7PHI5YKC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.03537&json=true","fetch_graph":"https://pith.science/api/pith-number/7PHI5YKCK3JHDMDUKP6VLU655V/graph.json","fetch_events":"https://pith.science/api/pith-number/7PHI5YKCK3JHDMDUKP6VLU655V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V/action/storage_attestation","attest_author":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V/action/author_attestation","sign_citation":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V/action/citation_signature","submit_replication":"https://pith.science/pith/7PHI5YKCK3JHDMDUKP6VLU655V/action/replication_record"}},"created_at":"2026-08-05T01:36:03.993818+00:00","updated_at":"2026-08-05T01:36:03.993818+00:00"}