{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:F5BAVY75G5N6XYNG4WPZU32KUF","short_pith_number":"pith:F5BAVY75","schema_version":"1.0","canonical_sha256":"2f420ae3fd375bebe1a6e59f9a6f4aa1495394d29de7d5f8e3d586aedf55f2e2","source":{"kind":"arxiv","id":"2605.30728","version":1},"attestation_state":"computed","paper":{"title":"Reducing the GPU Memory Bottleneck with Lossless Compression for ML -- Extended","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.LG","authors_text":"Aditya K Kamath, Arvind Krishnamurthy, Marco Canini, Simon Peter","submitted_at":"2026-05-29T01:45:28Z","abstract_excerpt":"Machine learning (ML) training and inference often process data sets far exceeding GPU memory capacity, forcing them to rely on PCIe for on-demand tensor transfers, causing critical transfer bottlenecks. Lossy compression has been proposed to relieve bottlenecks but introduces workload-dependent accuracy loss, making it complex or even prohibitive to use in existing ML deployments. We explore lossless compression as an alternative that avoids this deployment complexity. We identify where lossless compression can be integrated into ML pipelines while minimizing interference with GPU execution. "},"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":"2605.30728","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T01:45:28Z","cross_cats_sorted":["cs.DC"],"title_canon_sha256":"6e659d9f1d3db1a5813890104f31294387ea11c79ba3407ef9e81bd3139bf06b","abstract_canon_sha256":"2ea81056ed13b2de3dae078aee4e81235692bbca359aeaa3820eb32ee5e94e19"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-01T01:03:12.659744Z","signature_b64":"AxDfg3YwGd5K7/6+YE7iXZz2r4PP3Ao69VlzA0+rGiIEnww6q01rmRFaZ6h6bC21d6w+htt5UhumLalSaUJkCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2f420ae3fd375bebe1a6e59f9a6f4aa1495394d29de7d5f8e3d586aedf55f2e2","last_reissued_at":"2026-06-01T01:03:12.658713Z","signature_status":"signed_v1","first_computed_at":"2026-06-01T01:03:12.658713Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reducing the GPU Memory Bottleneck with Lossless Compression for ML -- Extended","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.LG","authors_text":"Aditya K Kamath, Arvind Krishnamurthy, Marco Canini, Simon Peter","submitted_at":"2026-05-29T01:45:28Z","abstract_excerpt":"Machine learning (ML) training and inference often process data sets far exceeding GPU memory capacity, forcing them to rely on PCIe for on-demand tensor transfers, causing critical transfer bottlenecks. Lossy compression has been proposed to relieve bottlenecks but introduces workload-dependent accuracy loss, making it complex or even prohibitive to use in existing ML deployments. We explore lossless compression as an alternative that avoids this deployment complexity. We identify where lossless compression can be integrated into ML pipelines while minimizing interference with GPU execution. "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.30728","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/2605.30728/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":"2605.30728","created_at":"2026-06-01T01:03:12.658886+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.30728v1","created_at":"2026-06-01T01:03:12.658886+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.30728","created_at":"2026-06-01T01:03:12.658886+00:00"},{"alias_kind":"pith_short_12","alias_value":"F5BAVY75G5N6","created_at":"2026-06-01T01:03:12.658886+00:00"},{"alias_kind":"pith_short_16","alias_value":"F5BAVY75G5N6XYNG","created_at":"2026-06-01T01:03:12.658886+00:00"},{"alias_kind":"pith_short_8","alias_value":"F5BAVY75","created_at":"2026-06-01T01:03:12.658886+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/F5BAVY75G5N6XYNG4WPZU32KUF","json":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF.json","graph_json":"https://pith.science/api/pith-number/F5BAVY75G5N6XYNG4WPZU32KUF/graph.json","events_json":"https://pith.science/api/pith-number/F5BAVY75G5N6XYNG4WPZU32KUF/events.json","paper":"https://pith.science/paper/F5BAVY75"},"agent_actions":{"view_html":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF","download_json":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF.json","view_paper":"https://pith.science/paper/F5BAVY75","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.30728&json=true","fetch_graph":"https://pith.science/api/pith-number/F5BAVY75G5N6XYNG4WPZU32KUF/graph.json","fetch_events":"https://pith.science/api/pith-number/F5BAVY75G5N6XYNG4WPZU32KUF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF/action/storage_attestation","attest_author":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF/action/author_attestation","sign_citation":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF/action/citation_signature","submit_replication":"https://pith.science/pith/F5BAVY75G5N6XYNG4WPZU32KUF/action/replication_record"}},"created_at":"2026-06-01T01:03:12.658886+00:00","updated_at":"2026-06-01T01:03:12.658886+00:00"}