{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LXHGMB5Q2VFZXAV4TYLGG5A7OP","short_pith_number":"pith:LXHGMB5Q","schema_version":"1.0","canonical_sha256":"5dce6607b0d54b9b82bc9e1663741f73e1ea6940d4a0570d8d07429715c2ade6","source":{"kind":"arxiv","id":"2512.10236","version":2},"attestation_state":"computed","paper":{"title":"Design Space Exploration of DMA based Finer-Grain Compute Communication Overlap","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AR","cs.LG"],"primary_cat":"cs.DC","authors_text":"Lizy K. John, Mahzabeen Islam, Shagnik Pal, Shaizeen Aga, Suchita Pati","submitted_at":"2025-12-11T02:43:27Z","abstract_excerpt":"Modern ML workloads demand distributing training and inference across multiple GPUs. However, these parallelization techniques often suffer from exposed critical-path communication, leaving a potential 1.7x speedup on the table through compute-communication overlap. Prior overlapping methods harness the fact that ML model state and inputs are already sharded into the number of GPUs, and overlap the compute and communication at shard granularity. However, such coarse-grained overlap suffers from limited network topology support, and suboptimal dataflows. In this work, we instead make a case for"},"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":"2512.10236","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-12-11T02:43:27Z","cross_cats_sorted":["cs.AR","cs.LG"],"title_canon_sha256":"9b4e28b5f8fa614e88a867fa9db600a6efafe6f9096f19559534ac4183c60def","abstract_canon_sha256":"769e436e97264ea75a7fc6bbfcfb0572bf485eaae4c1a57190a01a904e8d0eb8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:39.677061Z","signature_b64":"NaA3MhyJAShQ3Gxaf6ycXxp1AI5V6qSjJ/SZ0GZv4DK8IH/DtaQXrQLHM7T0r+v9HQZG8t9RILQhE1QqFz1FCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5dce6607b0d54b9b82bc9e1663741f73e1ea6940d4a0570d8d07429715c2ade6","last_reissued_at":"2026-06-02T01:03:39.676576Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:39.676576Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Design Space Exploration of DMA based Finer-Grain Compute Communication Overlap","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AR","cs.LG"],"primary_cat":"cs.DC","authors_text":"Lizy K. John, Mahzabeen Islam, Shagnik Pal, Shaizeen Aga, Suchita Pati","submitted_at":"2025-12-11T02:43:27Z","abstract_excerpt":"Modern ML workloads demand distributing training and inference across multiple GPUs. However, these parallelization techniques often suffer from exposed critical-path communication, leaving a potential 1.7x speedup on the table through compute-communication overlap. Prior overlapping methods harness the fact that ML model state and inputs are already sharded into the number of GPUs, and overlap the compute and communication at shard granularity. However, such coarse-grained overlap suffers from limited network topology support, and suboptimal dataflows. In this work, we instead make a case for"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.10236","kind":"arxiv","version":2},"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/2512.10236/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":"2512.10236","created_at":"2026-06-02T01:03:39.676641+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.10236v2","created_at":"2026-06-02T01:03:39.676641+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.10236","created_at":"2026-06-02T01:03:39.676641+00:00"},{"alias_kind":"pith_short_12","alias_value":"LXHGMB5Q2VFZ","created_at":"2026-06-02T01:03:39.676641+00:00"},{"alias_kind":"pith_short_16","alias_value":"LXHGMB5Q2VFZXAV4","created_at":"2026-06-02T01:03:39.676641+00:00"},{"alias_kind":"pith_short_8","alias_value":"LXHGMB5Q","created_at":"2026-06-02T01:03:39.676641+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/LXHGMB5Q2VFZXAV4TYLGG5A7OP","json":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP.json","graph_json":"https://pith.science/api/pith-number/LXHGMB5Q2VFZXAV4TYLGG5A7OP/graph.json","events_json":"https://pith.science/api/pith-number/LXHGMB5Q2VFZXAV4TYLGG5A7OP/events.json","paper":"https://pith.science/paper/LXHGMB5Q"},"agent_actions":{"view_html":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP","download_json":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP.json","view_paper":"https://pith.science/paper/LXHGMB5Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.10236&json=true","fetch_graph":"https://pith.science/api/pith-number/LXHGMB5Q2VFZXAV4TYLGG5A7OP/graph.json","fetch_events":"https://pith.science/api/pith-number/LXHGMB5Q2VFZXAV4TYLGG5A7OP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP/action/storage_attestation","attest_author":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP/action/author_attestation","sign_citation":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP/action/citation_signature","submit_replication":"https://pith.science/pith/LXHGMB5Q2VFZXAV4TYLGG5A7OP/action/replication_record"}},"created_at":"2026-06-02T01:03:39.676641+00:00","updated_at":"2026-06-02T01:03:39.676641+00:00"}