{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:VGAJPDPUCR4AR4VSELPLW3WEC5","short_pith_number":"pith:VGAJPDPU","schema_version":"1.0","canonical_sha256":"a980978df4147808f2b222debb6ec41767c562f880ed791d855891f9c04237ca","source":{"kind":"arxiv","id":"2304.01433","version":3},"attestation_state":"computed","paper":{"title":"TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.PF"],"primary_cat":"cs.AR","authors_text":"Andy Swing, Brian Towles, Cliff Young, David Patterson, George Kurian, Lifeng Nai, Nishant Patil, Norman P. Jouppi, Peter Ma, Rahul Nagarajan, Sheng Li, Suvinay Subramanian, Xiang Zhou, Zongwei Zhou","submitted_at":"2023-04-04T00:52:46Z","abstract_excerpt":"In response to innovations in machine learning (ML) models, production workloads changed radically and rapidly. TPU v4 is the fifth Google domain specific architecture (DSA) and its third supercomputer for such ML models. Optical circuit switches (OCSes) dynamically reconfigure its interconnect topology to improve scale, availability, utilization, modularity, deployment, security, power, and performance; users can pick a twisted 3D torus topology if desired. Much cheaper, lower power, and faster than Infiniband, OCSes and underlying optical components are <5% of system cost and <3% of system p"},"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":"2304.01433","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2023-04-04T00:52:46Z","cross_cats_sorted":["cs.AI","cs.LG","cs.PF"],"title_canon_sha256":"1d6aa874ae7efbf2b884961b72528c45a6e283aa556a84e29d5d48e4230c54b4","abstract_canon_sha256":"9606c2fe4bdb92359d8969d0d0c6c973a164412c12368145a0e51279b032b925"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:03:11.875487Z","signature_b64":"SnUv8u3K6M5PtVqupvCUOuwHLFjarp2bpQE9+sFyT7GCEUSwoAGbHyHNnvaorRkhVObF536ycNEcOoDi53xjCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a980978df4147808f2b222debb6ec41767c562f880ed791d855891f9c04237ca","last_reissued_at":"2026-07-05T06:03:11.874990Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:03:11.874990Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.PF"],"primary_cat":"cs.AR","authors_text":"Andy Swing, Brian Towles, Cliff Young, David Patterson, George Kurian, Lifeng Nai, Nishant Patil, Norman P. Jouppi, Peter Ma, Rahul Nagarajan, Sheng Li, Suvinay Subramanian, Xiang Zhou, Zongwei Zhou","submitted_at":"2023-04-04T00:52:46Z","abstract_excerpt":"In response to innovations in machine learning (ML) models, production workloads changed radically and rapidly. TPU v4 is the fifth Google domain specific architecture (DSA) and its third supercomputer for such ML models. Optical circuit switches (OCSes) dynamically reconfigure its interconnect topology to improve scale, availability, utilization, modularity, deployment, security, power, and performance; users can pick a twisted 3D torus topology if desired. Much cheaper, lower power, and faster than Infiniband, OCSes and underlying optical components are <5% of system cost and <3% of system p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.01433","kind":"arxiv","version":3},"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/2304.01433/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":"2304.01433","created_at":"2026-07-05T06:03:11.875046+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.01433v3","created_at":"2026-07-05T06:03:11.875046+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.01433","created_at":"2026-07-05T06:03:11.875046+00:00"},{"alias_kind":"pith_short_12","alias_value":"VGAJPDPUCR4A","created_at":"2026-07-05T06:03:11.875046+00:00"},{"alias_kind":"pith_short_16","alias_value":"VGAJPDPUCR4AR4VS","created_at":"2026-07-05T06:03:11.875046+00:00"},{"alias_kind":"pith_short_8","alias_value":"VGAJPDPU","created_at":"2026-07-05T06:03:11.875046+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17566","citing_title":"AoiZora: Topology-Aware Auto-Parallel Optimization for Inference of Diffusion Transformers","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02100","citing_title":"PortBERT: Navigating the Depths of Portuguese Language Models","ref_index":87,"is_internal_anchor":false},{"citing_arxiv_id":"2308.03958","citing_title":"Simple synthetic data reduces sycophancy in large language models","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2502.10517","citing_title":"KernelBench: Can LLMs Write Efficient GPU Kernels?","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2603.28239","citing_title":"A Switch-Centric In-Network Architecture for Accelerating LLM Inference in Shared-Memory Network","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17808","citing_title":"Enabling AI ASICs for Zero Knowledge Proof","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10403","citing_title":"Latent Instruction Representation Alignment: defending against jailbreaks, backdoors and undesired knowledge in LLMs","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19286","citing_title":"Mass Matrix Assembly on Tensor Cores for Implicit Particle-In-Cell Methods","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5","json":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5.json","graph_json":"https://pith.science/api/pith-number/VGAJPDPUCR4AR4VSELPLW3WEC5/graph.json","events_json":"https://pith.science/api/pith-number/VGAJPDPUCR4AR4VSELPLW3WEC5/events.json","paper":"https://pith.science/paper/VGAJPDPU"},"agent_actions":{"view_html":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5","download_json":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5.json","view_paper":"https://pith.science/paper/VGAJPDPU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.01433&json=true","fetch_graph":"https://pith.science/api/pith-number/VGAJPDPUCR4AR4VSELPLW3WEC5/graph.json","fetch_events":"https://pith.science/api/pith-number/VGAJPDPUCR4AR4VSELPLW3WEC5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5/action/storage_attestation","attest_author":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5/action/author_attestation","sign_citation":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5/action/citation_signature","submit_replication":"https://pith.science/pith/VGAJPDPUCR4AR4VSELPLW3WEC5/action/replication_record"}},"created_at":"2026-07-05T06:03:11.875046+00:00","updated_at":"2026-07-05T06:03:11.875046+00:00"}