{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:IXE3WPGLEN62LQCG64FB4HAMON","short_pith_number":"pith:IXE3WPGL","canonical_record":{"source":{"id":"2512.07004","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MS","submitted_at":"2025-12-07T21:13:18Z","cross_cats_sorted":["cs.AR","cs.NA","math.NA"],"title_canon_sha256":"763583d8a8672cc94b3419c74c4108486c210a0403cc8e1663a704af35580479","abstract_canon_sha256":"83e1ff4c1bce5e25781f67420edc6dbf6faf83eb46e0e4cfd0fb32fdcdbe9bbc"},"schema_version":"1.0"},"canonical_sha256":"45c9bb3ccb237da5c046f70a1e1c0c7346b970c4e6313be223cdd9033f737576","source":{"kind":"arxiv","id":"2512.07004","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.07004","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"arxiv_version","alias_value":"2512.07004v4","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.07004","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"pith_short_12","alias_value":"IXE3WPGLEN62","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"pith_short_16","alias_value":"IXE3WPGLEN62LQCG","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"pith_short_8","alias_value":"IXE3WPGL","created_at":"2026-06-12T01:10:19Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:IXE3WPGLEN62LQCG64FB4HAMON","target":"record","payload":{"canonical_record":{"source":{"id":"2512.07004","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MS","submitted_at":"2025-12-07T21:13:18Z","cross_cats_sorted":["cs.AR","cs.NA","math.NA"],"title_canon_sha256":"763583d8a8672cc94b3419c74c4108486c210a0403cc8e1663a704af35580479","abstract_canon_sha256":"83e1ff4c1bce5e25781f67420edc6dbf6faf83eb46e0e4cfd0fb32fdcdbe9bbc"},"schema_version":"1.0"},"canonical_sha256":"45c9bb3ccb237da5c046f70a1e1c0c7346b970c4e6313be223cdd9033f737576","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-12T01:10:19.788113Z","signature_b64":"mMaC1DU0zMzRK2jVWPD+D1r6gyhTkLkmXkjMcdSle8k8yiy/dy6OVn10Qw3JQq4RPmzxARFRg8ug7HeRk6kVCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45c9bb3ccb237da5c046f70a1e1c0c7346b970c4e6313be223cdd9033f737576","last_reissued_at":"2026-06-12T01:10:19.787268Z","signature_status":"signed_v1","first_computed_at":"2026-06-12T01:10:19.787268Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2512.07004","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-12T01:10:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zXr01OZKfLMj9xNjNOVhIByk7kQWm54VxaKiHf/rN+X2fGKc2QXgdJoJq/g4TTi+kXhAe0LntGQNJFXu589lDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T16:25:58.596744Z"},"content_sha256":"9575cae035654ed5de015eed59c1156562671ea1b2e97fcca14d8b708c457597","schema_version":"1.0","event_id":"sha256:9575cae035654ed5de015eed59c1156562671ea1b2e97fcca14d8b708c457597"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:IXE3WPGLEN62LQCG64FB4HAMON","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Accurate Models of NVIDIA Tensor Cores","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Software models emulate the inner-product behavior of NVIDIA Tensor Cores for 8-, 16-, and 19-bit formats across V100 to B200 GPUs.","cross_cats":["cs.AR","cs.NA","math.NA"],"primary_cat":"cs.MS","authors_text":"Faizan A. Khattak, Mantas Mikaitis","submitted_at":"2025-12-07T21:13:18Z","abstract_excerpt":"Matrix multiplication is a fundamental operation in both training of neural networks and inference. To accelerate matrix multiplication, Graphical Processing Units (GPUs) provide it implemented in hardware. Due to the increased throughput over the software-based matrix multiplication, the multipliers are increasingly used outside of AI, to accelerate various applications in scientific computing. However, matrix multipliers targeted at AI are at present not compliant with IEEE 754 floating-point arithmetic behaviour, with different vendors offering different numerical features. This leads to no"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We present software models for emulating the inner product behaviour of low- and mixed-precision matrix multipliers in the V100, A100, H100 and B200 data center GPUs in most supported input formats of interest to mixed-precision algorithm developers: 8-, 16-, and 19-bit floating point.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The chosen test vectors are sufficient to distinguish the numerical features (rounding, accumulator width, normalization, carry bits) of each hardware generation and remain reliable when applied to new platforms.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Software models accurately emulate NVIDIA Tensor Core inner-product behavior in low- and mixed-precision formats across V100, A100, H100, and B200 GPUs.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Software models emulate the inner-product behavior of NVIDIA Tensor Cores for 8-, 16-, and 19-bit formats across V100 to B200 GPUs.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"03aa984a15cad6e12a4223599463b3afab33e2d82e16b28cc94552cd368e4a79"},"source":{"id":"2512.07004","kind":"arxiv","version":4},"verdict":{"id":"570d2853-97d2-46ed-8896-a07ff47f5452","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T00:49:23.385742Z","strongest_claim":"We present software models for emulating the inner product behaviour of low- and mixed-precision matrix multipliers in the V100, A100, H100 and B200 data center GPUs in most supported input formats of interest to mixed-precision algorithm developers: 8-, 16-, and 19-bit floating point.","one_line_summary":"Software models accurately emulate NVIDIA Tensor Core inner-product behavior in low- and mixed-precision formats across V100, A100, H100, and B200 GPUs.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The chosen test vectors are sufficient to distinguish the numerical features (rounding, accumulator width, normalization, carry bits) of each hardware generation and remain reliable when applied to new platforms.","pith_extraction_headline":"Software models emulate the inner-product behavior of NVIDIA Tensor Cores for 8-, 16-, and 19-bit formats across V100 to B200 GPUs."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2512.07004/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"},"verdict_id":"570d2853-97d2-46ed-8896-a07ff47f5452"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-12T01:10:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lP6ibywop5alyUPtGVpngC3sCq+KIS91qpmWbTgJ6k4DU7XBSXsyLrJXMJ5jLzUszqpRqK4yH+E+xAg40KWhAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T16:25:58.598304Z"},"content_sha256":"413ea04de4b0099128dc9576e88a1b6bdceb59818d347f06de382b1ab0ca5d66","schema_version":"1.0","event_id":"sha256:413ea04de4b0099128dc9576e88a1b6bdceb59818d347f06de382b1ab0ca5d66"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IXE3WPGLEN62LQCG64FB4HAMON/bundle.json","state_url":"https://pith.science/pith/IXE3WPGLEN62LQCG64FB4HAMON/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IXE3WPGLEN62LQCG64FB4HAMON/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-03T16:25:58Z","links":{"resolver":"https://pith.science/pith/IXE3WPGLEN62LQCG64FB4HAMON","bundle":"https://pith.science/pith/IXE3WPGLEN62LQCG64FB4HAMON/bundle.json","state":"https://pith.science/pith/IXE3WPGLEN62LQCG64FB4HAMON/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IXE3WPGLEN62LQCG64FB4HAMON/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:IXE3WPGLEN62LQCG64FB4HAMON","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"83e1ff4c1bce5e25781f67420edc6dbf6faf83eb46e0e4cfd0fb32fdcdbe9bbc","cross_cats_sorted":["cs.AR","cs.NA","math.NA"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MS","submitted_at":"2025-12-07T21:13:18Z","title_canon_sha256":"763583d8a8672cc94b3419c74c4108486c210a0403cc8e1663a704af35580479"},"schema_version":"1.0","source":{"id":"2512.07004","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.07004","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"arxiv_version","alias_value":"2512.07004v4","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.07004","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"pith_short_12","alias_value":"IXE3WPGLEN62","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"pith_short_16","alias_value":"IXE3WPGLEN62LQCG","created_at":"2026-06-12T01:10:19Z"},{"alias_kind":"pith_short_8","alias_value":"IXE3WPGL","created_at":"2026-06-12T01:10:19Z"}],"graph_snapshots":[{"event_id":"sha256:413ea04de4b0099128dc9576e88a1b6bdceb59818d347f06de382b1ab0ca5d66","target":"graph","created_at":"2026-06-12T01:10:19Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We present software models for emulating the inner product behaviour of low- and mixed-precision matrix multipliers in the V100, A100, H100 and B200 data center GPUs in most supported input formats of interest to mixed-precision algorithm developers: 8-, 16-, and 19-bit floating point."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The chosen test vectors are sufficient to distinguish the numerical features (rounding, accumulator width, normalization, carry bits) of each hardware generation and remain reliable when applied to new platforms."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Software models accurately emulate NVIDIA Tensor Core inner-product behavior in low- and mixed-precision formats across V100, A100, H100, and B200 GPUs."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Software models emulate the inner-product behavior of NVIDIA Tensor Cores for 8-, 16-, and 19-bit formats across V100 to B200 GPUs."}],"snapshot_sha256":"03aa984a15cad6e12a4223599463b3afab33e2d82e16b28cc94552cd368e4a79"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2512.07004/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Matrix multiplication is a fundamental operation in both training of neural networks and inference. To accelerate matrix multiplication, Graphical Processing Units (GPUs) provide it implemented in hardware. Due to the increased throughput over the software-based matrix multiplication, the multipliers are increasingly used outside of AI, to accelerate various applications in scientific computing. However, matrix multipliers targeted at AI are at present not compliant with IEEE 754 floating-point arithmetic behaviour, with different vendors offering different numerical features. This leads to no","authors_text":"Faizan A. Khattak, Mantas Mikaitis","cross_cats":["cs.AR","cs.NA","math.NA"],"headline":"Software models emulate the inner-product behavior of NVIDIA Tensor Cores for 8-, 16-, and 19-bit formats across V100 to B200 GPUs.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MS","submitted_at":"2025-12-07T21:13:18Z","title":"Accurate Models of NVIDIA Tensor Cores"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.07004","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-17T00:49:23.385742Z","id":"570d2853-97d2-46ed-8896-a07ff47f5452","model_set":{"reader":"grok-4.3"},"one_line_summary":"Software models accurately emulate NVIDIA Tensor Core inner-product behavior in low- and mixed-precision formats across V100, A100, H100, and B200 GPUs.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Software models emulate the inner-product behavior of NVIDIA Tensor Cores for 8-, 16-, and 19-bit formats across V100 to B200 GPUs.","strongest_claim":"We present software models for emulating the inner product behaviour of low- and mixed-precision matrix multipliers in the V100, A100, H100 and B200 data center GPUs in most supported input formats of interest to mixed-precision algorithm developers: 8-, 16-, and 19-bit floating point.","weakest_assumption":"The chosen test vectors are sufficient to distinguish the numerical features (rounding, accumulator width, normalization, carry bits) of each hardware generation and remain reliable when applied to new platforms."}},"verdict_id":"570d2853-97d2-46ed-8896-a07ff47f5452"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:9575cae035654ed5de015eed59c1156562671ea1b2e97fcca14d8b708c457597","target":"record","created_at":"2026-06-12T01:10:19Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"83e1ff4c1bce5e25781f67420edc6dbf6faf83eb46e0e4cfd0fb32fdcdbe9bbc","cross_cats_sorted":["cs.AR","cs.NA","math.NA"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MS","submitted_at":"2025-12-07T21:13:18Z","title_canon_sha256":"763583d8a8672cc94b3419c74c4108486c210a0403cc8e1663a704af35580479"},"schema_version":"1.0","source":{"id":"2512.07004","kind":"arxiv","version":4}},"canonical_sha256":"45c9bb3ccb237da5c046f70a1e1c0c7346b970c4e6313be223cdd9033f737576","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"45c9bb3ccb237da5c046f70a1e1c0c7346b970c4e6313be223cdd9033f737576","first_computed_at":"2026-06-12T01:10:19.787268Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-12T01:10:19.787268Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"mMaC1DU0zMzRK2jVWPD+D1r6gyhTkLkmXkjMcdSle8k8yiy/dy6OVn10Qw3JQq4RPmzxARFRg8ug7HeRk6kVCA==","signature_status":"signed_v1","signed_at":"2026-06-12T01:10:19.788113Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.07004","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9575cae035654ed5de015eed59c1156562671ea1b2e97fcca14d8b708c457597","sha256:413ea04de4b0099128dc9576e88a1b6bdceb59818d347f06de382b1ab0ca5d66"],"state_sha256":"4a90b11c1bfc927cadbe5ae1056b537b8edbf9321657316136113662b6741cb9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ImrYNWnA035sLEMQSzMuEkUifLI5rcVQwDYHaykPAo+azgomHvoUOApOZP69LnC8TGzgehAx5mmGw9gRNjacBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-03T16:25:58.607492Z","bundle_sha256":"e0fb3b1fd6b8cba778979773df973545a364c7f41ab8da73c3ebe8832167cc92"}}