{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CYQSKMLJ7Q44Z3VKXBMP5OAESZ","short_pith_number":"pith:CYQSKMLJ","schema_version":"1.0","canonical_sha256":"1621253169fc39cceeaab858feb804964e72530ccc0e11b6c566073906b1fa19","source":{"kind":"arxiv","id":"2606.01927","version":1},"attestation_state":"computed","paper":{"title":"Scaling LLM Inference Beyond Amdahl`s Limits via Eliminating Non-Scalable Overheads","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Alan Zhao, Cyril Y. He, Wei Xu","submitted_at":"2026-06-01T08:58:23Z","abstract_excerpt":"Deployers of online LLM services usually seek to maximize cluster-wide performance given a fixed number of GPUs. Tensor parallelism (TP) is necessary to fit modern models but scales sub-linearly as the TP degree t grows, due to cross-GPU communication and non-scalable runtime work, as predicted by Amdahl's Law. Conversely, increasing t improves memory efficiency and alleviates KV-cache contention and swapping. We identify and validate an empirical optimal TP degree t_e that balances these effects. We present Albireo, a parallel inference system that raises the attainable t_e by shrinking the n"},"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":"2606.01927","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.DC","submitted_at":"2026-06-01T08:58:23Z","cross_cats_sorted":[],"title_canon_sha256":"240fb490009746a9e7f76983c8e6e9e308c6d738c8431cf64a131e846f093cfb","abstract_canon_sha256":"1d33634277a676123bf0abc09f0f5821ab1b891ea8f0bfe24d9e7fe4d255f3c5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:05:00.749388Z","signature_b64":"cogRO6d4FWfyOjDpkYI/Xu+GsFUCixhbeD8qqIvRb3pmOoYLKi8D37dwLoHsI11fpNefxU5hA3u255hwwDYhCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1621253169fc39cceeaab858feb804964e72530ccc0e11b6c566073906b1fa19","last_reissued_at":"2026-06-02T02:05:00.748980Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:05:00.748980Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling LLM Inference Beyond Amdahl`s Limits via Eliminating Non-Scalable Overheads","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Alan Zhao, Cyril Y. He, Wei Xu","submitted_at":"2026-06-01T08:58:23Z","abstract_excerpt":"Deployers of online LLM services usually seek to maximize cluster-wide performance given a fixed number of GPUs. Tensor parallelism (TP) is necessary to fit modern models but scales sub-linearly as the TP degree t grows, due to cross-GPU communication and non-scalable runtime work, as predicted by Amdahl's Law. Conversely, increasing t improves memory efficiency and alleviates KV-cache contention and swapping. We identify and validate an empirical optimal TP degree t_e that balances these effects. We present Albireo, a parallel inference system that raises the attainable t_e by shrinking the n"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01927","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/2606.01927/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":"2606.01927","created_at":"2026-06-02T02:05:00.749033+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01927v1","created_at":"2026-06-02T02:05:00.749033+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01927","created_at":"2026-06-02T02:05:00.749033+00:00"},{"alias_kind":"pith_short_12","alias_value":"CYQSKMLJ7Q44","created_at":"2026-06-02T02:05:00.749033+00:00"},{"alias_kind":"pith_short_16","alias_value":"CYQSKMLJ7Q44Z3VK","created_at":"2026-06-02T02:05:00.749033+00:00"},{"alias_kind":"pith_short_8","alias_value":"CYQSKMLJ","created_at":"2026-06-02T02:05:00.749033+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/CYQSKMLJ7Q44Z3VKXBMP5OAESZ","json":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ.json","graph_json":"https://pith.science/api/pith-number/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/graph.json","events_json":"https://pith.science/api/pith-number/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/events.json","paper":"https://pith.science/paper/CYQSKMLJ"},"agent_actions":{"view_html":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ","download_json":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ.json","view_paper":"https://pith.science/paper/CYQSKMLJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01927&json=true","fetch_graph":"https://pith.science/api/pith-number/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/graph.json","fetch_events":"https://pith.science/api/pith-number/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/action/storage_attestation","attest_author":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/action/author_attestation","sign_citation":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/action/citation_signature","submit_replication":"https://pith.science/pith/CYQSKMLJ7Q44Z3VKXBMP5OAESZ/action/replication_record"}},"created_at":"2026-06-02T02:05:00.749033+00:00","updated_at":"2026-06-02T02:05:00.749033+00:00"}