{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:TGPSSDHLQQZXCIQHDN6TA5UTNQ","short_pith_number":"pith:TGPSSDHL","schema_version":"1.0","canonical_sha256":"999f290ceb84337122071b7d3076936c09407561bd6f7a70ef2d89dd4428d80d","source":{"kind":"arxiv","id":"2411.01783","version":3},"attestation_state":"computed","paper":{"title":"Context Parallelism for Scalable Million-Token Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.DC","authors_text":"Amy Yang, Aya Ibrahim, Bangsheng Tang, Grigory Sizov, Jeremy Reizenstein, Jianyu Huang, Jingyi Yang, Jongsoo Park, Xinfeng Xie","submitted_at":"2024-11-04T04:15:36Z","abstract_excerpt":"We present context parallelism for long-context large language model inference, which achieves near-linear scaling for long-context prefill latency with up to 128 H100 GPUs across 16 nodes. Particularly, our method achieves 1M context prefill with Llama3 405B model in 77s (93% parallelization efficiency, 63% FLOPS utilization) and 128K context prefill in 3.8s. We develop two lossless exact ring attention variants: pass-KV and pass-Q to cover a wide range of use cases with the state-of-the-art performance: full prefill, persistent KV prefill and decode. Benchmarks on H100 GPU hosts inter-connec"},"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":"2411.01783","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2024-11-04T04:15:36Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"66d84f63683bd4ad312b0c106e4702ff532eb6ca8a58cc31907910ee0f09d91f","abstract_canon_sha256":"46fe8828dc7937ec4b598c842259430d355a384666314b65dd1fe0b71c004a4e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:51:27.857409Z","signature_b64":"mPuGAbGNA+s8A6Qz+wtTWPKqaaZdedcaiKPhDI+WS4WD9cxJpazhiK8eGhdlaNhgNSozxho45/LtBphXtFknBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"999f290ceb84337122071b7d3076936c09407561bd6f7a70ef2d89dd4428d80d","last_reissued_at":"2026-07-05T10:51:27.856924Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:51:27.856924Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Context Parallelism for Scalable Million-Token Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.DC","authors_text":"Amy Yang, Aya Ibrahim, Bangsheng Tang, Grigory Sizov, Jeremy Reizenstein, Jianyu Huang, Jingyi Yang, Jongsoo Park, Xinfeng Xie","submitted_at":"2024-11-04T04:15:36Z","abstract_excerpt":"We present context parallelism for long-context large language model inference, which achieves near-linear scaling for long-context prefill latency with up to 128 H100 GPUs across 16 nodes. Particularly, our method achieves 1M context prefill with Llama3 405B model in 77s (93% parallelization efficiency, 63% FLOPS utilization) and 128K context prefill in 3.8s. We develop two lossless exact ring attention variants: pass-KV and pass-Q to cover a wide range of use cases with the state-of-the-art performance: full prefill, persistent KV prefill and decode. Benchmarks on H100 GPU hosts inter-connec"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.01783","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/2411.01783/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":"2411.01783","created_at":"2026-07-05T10:51:27.856978+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.01783v3","created_at":"2026-07-05T10:51:27.856978+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.01783","created_at":"2026-07-05T10:51:27.856978+00:00"},{"alias_kind":"pith_short_12","alias_value":"TGPSSDHLQQZX","created_at":"2026-07-05T10:51:27.856978+00:00"},{"alias_kind":"pith_short_16","alias_value":"TGPSSDHLQQZXCIQH","created_at":"2026-07-05T10:51:27.856978+00:00"},{"alias_kind":"pith_short_8","alias_value":"TGPSSDHL","created_at":"2026-07-05T10:51:27.856978+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10493","citing_title":"Achieving Cloud-Grade SLOs for Local Mixture-of-Experts Inference through CPU-GPU Hybrid Design","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01502","citing_title":"Move the Query, Not the Cache: Characterizing Cross-Instance Latent Attention Redistribution Across GPU Fabrics","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17059","citing_title":"Towards Distributed Inference of LLMs on a P2P Network","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29207","citing_title":"KernelFlume: Elastic Core-Attention Scaling for Agentic Long-Context Decoding","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2510.15596","citing_title":"PRISM: Probabilistic Runtime Insights and Scalable Performance Modeling for Large-Scale Distributed Training","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00831","citing_title":"GhostServe: A Lightweight Checkpointing System in the Shadow for Fault-Tolerant LLM Serving","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08542","citing_title":"Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstruction","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15732","citing_title":"Accuracy Is Speed: Towards Long-Context-Aware Routing for Distributed LLM Serving","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ","json":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ.json","graph_json":"https://pith.science/api/pith-number/TGPSSDHLQQZXCIQHDN6TA5UTNQ/graph.json","events_json":"https://pith.science/api/pith-number/TGPSSDHLQQZXCIQHDN6TA5UTNQ/events.json","paper":"https://pith.science/paper/TGPSSDHL"},"agent_actions":{"view_html":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ","download_json":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ.json","view_paper":"https://pith.science/paper/TGPSSDHL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.01783&json=true","fetch_graph":"https://pith.science/api/pith-number/TGPSSDHLQQZXCIQHDN6TA5UTNQ/graph.json","fetch_events":"https://pith.science/api/pith-number/TGPSSDHLQQZXCIQHDN6TA5UTNQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ/action/storage_attestation","attest_author":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ/action/author_attestation","sign_citation":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ/action/citation_signature","submit_replication":"https://pith.science/pith/TGPSSDHLQQZXCIQHDN6TA5UTNQ/action/replication_record"}},"created_at":"2026-07-05T10:51:27.856978+00:00","updated_at":"2026-07-05T10:51:27.856978+00:00"}