{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:NDHSG6IIORM57SA6WMAPCTNM2U","short_pith_number":"pith:NDHSG6II","canonical_record":{"source":{"id":"2502.10659","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2025-02-15T03:56:22Z","cross_cats_sorted":[],"title_canon_sha256":"0851399bfbc3a4afa3c5a2ac6080ad9c0c6d690ea74bc0444f8251cd52d950d8","abstract_canon_sha256":"e1e71146ce597e3d6224bf52fb8b8c08a9ff029d7de97c8188ca1761ed6160dd"},"schema_version":"1.0"},"canonical_sha256":"68cf2379087459dfc81eb300f14dacd51a6214bda67a30fe9fd5191e8bc82052","source":{"kind":"arxiv","id":"2502.10659","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.10659","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"arxiv_version","alias_value":"2502.10659v1","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.10659","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"pith_short_12","alias_value":"NDHSG6IIORM5","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"pith_short_16","alias_value":"NDHSG6IIORM57SA6","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"pith_short_8","alias_value":"NDHSG6II","created_at":"2026-07-05T10:14:55Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:NDHSG6IIORM57SA6WMAPCTNM2U","target":"record","payload":{"canonical_record":{"source":{"id":"2502.10659","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2025-02-15T03:56:22Z","cross_cats_sorted":[],"title_canon_sha256":"0851399bfbc3a4afa3c5a2ac6080ad9c0c6d690ea74bc0444f8251cd52d950d8","abstract_canon_sha256":"e1e71146ce597e3d6224bf52fb8b8c08a9ff029d7de97c8188ca1761ed6160dd"},"schema_version":"1.0"},"canonical_sha256":"68cf2379087459dfc81eb300f14dacd51a6214bda67a30fe9fd5191e8bc82052","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:14:55.897743Z","signature_b64":"O+ID00nijxlUUbA2uQ5YElD3ABy8DUtiNPAhIG/+HNzpgNPtGVqKQ4j5/fUF+uAOKJ1+LsZI3N1pwnKxShkBAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"68cf2379087459dfc81eb300f14dacd51a6214bda67a30fe9fd5191e8bc82052","last_reissued_at":"2026-07-05T10:14:55.897239Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:14:55.897239Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.10659","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:14:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vR01+KXk/KSfa5ix/cLg7Hian4HbbQ02gao6djzZI3LM7Js/ix1EHo0y/Ve/FiD1mAh2KwerXmw9doiKPLvuCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T07:35:57.695686Z"},"content_sha256":"9b44f1f02a2016f8a4861ddc750502f4db5d497ccc09c69ef4aab8789629e1cf","schema_version":"1.0","event_id":"sha256:9b44f1f02a2016f8a4861ddc750502f4db5d497ccc09c69ef4aab8789629e1cf"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:NDHSG6IIORM57SA6WMAPCTNM2U","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Pushing up to the Limit of Memory Bandwidth and Capacity Utilization for Efficient LLM Decoding on Embedded FPGA","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AR","authors_text":"Dongcheng Zhao, Guobin Shen, Jindong Li, Qian Zhang, Tenglong Li, Yi Zeng","submitted_at":"2025-02-15T03:56:22Z","abstract_excerpt":"The extremely high computational and storage demands of large language models have excluded most edge devices, which were widely used for efficient machine learning, from being viable options. A typical edge device usually only has 4GB of memory capacity and a bandwidth of less than 20GB/s, while a large language model quantized to 4-bit precision with 7B parameters already requires 3.5GB of capacity, and its decoding process is purely bandwidth-bound. In this paper, we aim to explore these limits by proposing a hardware accelerator for large language model (LLM) inference on the Zynq-based KV"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.10659","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/2502.10659/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":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:14:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"bWEjd31PWgiPIb9n4R2imVEARaGIbwg7qP0jSGzRUps5Qx6adPtRsf0gx4TpEwYeVbvSMEVzTdoYOnJI9T71Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T07:35:57.696230Z"},"content_sha256":"4b33e007a1ca3d053bed7f2409b145f0c5eeb0deaf0485ef5d5fe1f20baa0bc8","schema_version":"1.0","event_id":"sha256:4b33e007a1ca3d053bed7f2409b145f0c5eeb0deaf0485ef5d5fe1f20baa0bc8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NDHSG6IIORM57SA6WMAPCTNM2U/bundle.json","state_url":"https://pith.science/pith/NDHSG6IIORM57SA6WMAPCTNM2U/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NDHSG6IIORM57SA6WMAPCTNM2U/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":"2026-08-08T07:35:57Z","links":{"resolver":"https://pith.science/pith/NDHSG6IIORM57SA6WMAPCTNM2U","bundle":"https://pith.science/pith/NDHSG6IIORM57SA6WMAPCTNM2U/bundle.json","state":"https://pith.science/pith/NDHSG6IIORM57SA6WMAPCTNM2U/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NDHSG6IIORM57SA6WMAPCTNM2U/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:NDHSG6IIORM57SA6WMAPCTNM2U","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":"e1e71146ce597e3d6224bf52fb8b8c08a9ff029d7de97c8188ca1761ed6160dd","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2025-02-15T03:56:22Z","title_canon_sha256":"0851399bfbc3a4afa3c5a2ac6080ad9c0c6d690ea74bc0444f8251cd52d950d8"},"schema_version":"1.0","source":{"id":"2502.10659","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.10659","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"arxiv_version","alias_value":"2502.10659v1","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.10659","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"pith_short_12","alias_value":"NDHSG6IIORM5","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"pith_short_16","alias_value":"NDHSG6IIORM57SA6","created_at":"2026-07-05T10:14:55Z"},{"alias_kind":"pith_short_8","alias_value":"NDHSG6II","created_at":"2026-07-05T10:14:55Z"}],"graph_snapshots":[{"event_id":"sha256:4b33e007a1ca3d053bed7f2409b145f0c5eeb0deaf0485ef5d5fe1f20baa0bc8","target":"graph","created_at":"2026-07-05T10:14:55Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2502.10659/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The extremely high computational and storage demands of large language models have excluded most edge devices, which were widely used for efficient machine learning, from being viable options. A typical edge device usually only has 4GB of memory capacity and a bandwidth of less than 20GB/s, while a large language model quantized to 4-bit precision with 7B parameters already requires 3.5GB of capacity, and its decoding process is purely bandwidth-bound. In this paper, we aim to explore these limits by proposing a hardware accelerator for large language model (LLM) inference on the Zynq-based KV","authors_text":"Dongcheng Zhao, Guobin Shen, Jindong Li, Qian Zhang, Tenglong Li, Yi Zeng","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2025-02-15T03:56:22Z","title":"Pushing up to the Limit of Memory Bandwidth and Capacity Utilization for Efficient LLM Decoding on Embedded FPGA"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.10659","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:9b44f1f02a2016f8a4861ddc750502f4db5d497ccc09c69ef4aab8789629e1cf","target":"record","created_at":"2026-07-05T10:14:55Z","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":"e1e71146ce597e3d6224bf52fb8b8c08a9ff029d7de97c8188ca1761ed6160dd","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AR","submitted_at":"2025-02-15T03:56:22Z","title_canon_sha256":"0851399bfbc3a4afa3c5a2ac6080ad9c0c6d690ea74bc0444f8251cd52d950d8"},"schema_version":"1.0","source":{"id":"2502.10659","kind":"arxiv","version":1}},"canonical_sha256":"68cf2379087459dfc81eb300f14dacd51a6214bda67a30fe9fd5191e8bc82052","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"68cf2379087459dfc81eb300f14dacd51a6214bda67a30fe9fd5191e8bc82052","first_computed_at":"2026-07-05T10:14:55.897239Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:14:55.897239Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"O+ID00nijxlUUbA2uQ5YElD3ABy8DUtiNPAhIG/+HNzpgNPtGVqKQ4j5/fUF+uAOKJ1+LsZI3N1pwnKxShkBAw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:14:55.897743Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.10659","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9b44f1f02a2016f8a4861ddc750502f4db5d497ccc09c69ef4aab8789629e1cf","sha256:4b33e007a1ca3d053bed7f2409b145f0c5eeb0deaf0485ef5d5fe1f20baa0bc8"],"state_sha256":"0d02e9828278f3c77af3cadb263f6c5b8b22ca1c192761d95a4fc9a4f964f387"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/6RMVNESd3aOKN0Ak8QnbW7pxS5xsOenQB3WotrZlJnDJSWA4JNI7nGvZxwtg+2W2u7+aggc83HG3hNncvZyDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T07:35:57.701148Z","bundle_sha256":"1f57683f31cc5a1e96136cb0032764c150d5fd183f8d847cc6afb8b8397c6871"}}