{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LGWZOFJYZEUUBFYR7VKZXVXFQK","short_pith_number":"pith:LGWZOFJY","schema_version":"1.0","canonical_sha256":"59ad971538c929409711fd559bd6e5829948fa89a0ebd421ff3b56455422671b","source":{"kind":"arxiv","id":"2506.06235","version":1},"attestation_state":"computed","paper":{"title":"Optimizing Cloud-to-GPU Throughput for Deep Learning With Earth Observation Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Akram Zaytar, Anthony Ortiz, Caleb Robinson, Gilles Hacheme, Girmaw Abebe Tadesse, Juan M Lavista Ferres, Rahul M Dodhia, Tammy Glazer","submitted_at":"2025-06-06T16:54:13Z","abstract_excerpt":"Training deep learning models on petabyte-scale Earth observation (EO) data requires separating compute resources from data storage. However, standard PyTorch data loaders cannot keep modern GPUs utilized when streaming GeoTIFF files directly from cloud storage. In this work, we benchmark GeoTIFF loading throughput from both cloud object storage and local SSD, systematically testing different loader configurations and data parameters. We focus on tile-aligned reads and worker thread pools, using Bayesian optimization to find optimal settings for each storage type. Our optimized configurations "},"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":"2506.06235","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-06-06T16:54:13Z","cross_cats_sorted":[],"title_canon_sha256":"beae87a0ee7d5ebc3ac89f8c8c10e4bce0e2184a9396ec381d71887c44176c2d","abstract_canon_sha256":"db3929dd3c79888eb5ab5905a65993ef34bd191800188b3ae3435e6124891436"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:27.807915Z","signature_b64":"PCQJjSGYn/xNLEjdefkFsvHEGUL3znsoO7Q+SUpoNQBWcPwjM28qZ9xmDAxPNZFqe5Qo8JnZsjn/Uj8girVQDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"59ad971538c929409711fd559bd6e5829948fa89a0ebd421ff3b56455422671b","last_reissued_at":"2026-07-05T11:17:27.807415Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:27.807415Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Optimizing Cloud-to-GPU Throughput for Deep Learning With Earth Observation Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Akram Zaytar, Anthony Ortiz, Caleb Robinson, Gilles Hacheme, Girmaw Abebe Tadesse, Juan M Lavista Ferres, Rahul M Dodhia, Tammy Glazer","submitted_at":"2025-06-06T16:54:13Z","abstract_excerpt":"Training deep learning models on petabyte-scale Earth observation (EO) data requires separating compute resources from data storage. However, standard PyTorch data loaders cannot keep modern GPUs utilized when streaming GeoTIFF files directly from cloud storage. In this work, we benchmark GeoTIFF loading throughput from both cloud object storage and local SSD, systematically testing different loader configurations and data parameters. We focus on tile-aligned reads and worker thread pools, using Bayesian optimization to find optimal settings for each storage type. Our optimized configurations "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.06235","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/2506.06235/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":"2506.06235","created_at":"2026-07-05T11:17:27.807481+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.06235v1","created_at":"2026-07-05T11:17:27.807481+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.06235","created_at":"2026-07-05T11:17:27.807481+00:00"},{"alias_kind":"pith_short_12","alias_value":"LGWZOFJYZEUU","created_at":"2026-07-05T11:17:27.807481+00:00"},{"alias_kind":"pith_short_16","alias_value":"LGWZOFJYZEUUBFYR","created_at":"2026-07-05T11:17:27.807481+00:00"},{"alias_kind":"pith_short_8","alias_value":"LGWZOFJY","created_at":"2026-07-05T11:17:27.807481+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/LGWZOFJYZEUUBFYR7VKZXVXFQK","json":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK.json","graph_json":"https://pith.science/api/pith-number/LGWZOFJYZEUUBFYR7VKZXVXFQK/graph.json","events_json":"https://pith.science/api/pith-number/LGWZOFJYZEUUBFYR7VKZXVXFQK/events.json","paper":"https://pith.science/paper/LGWZOFJY"},"agent_actions":{"view_html":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK","download_json":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK.json","view_paper":"https://pith.science/paper/LGWZOFJY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.06235&json=true","fetch_graph":"https://pith.science/api/pith-number/LGWZOFJYZEUUBFYR7VKZXVXFQK/graph.json","fetch_events":"https://pith.science/api/pith-number/LGWZOFJYZEUUBFYR7VKZXVXFQK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK/action/storage_attestation","attest_author":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK/action/author_attestation","sign_citation":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK/action/citation_signature","submit_replication":"https://pith.science/pith/LGWZOFJYZEUUBFYR7VKZXVXFQK/action/replication_record"}},"created_at":"2026-07-05T11:17:27.807481+00:00","updated_at":"2026-07-05T11:17:27.807481+00:00"}