Unverified50% confidenceFactExact time
Continuous batching was first formally proposed by the Orca system from UC Berkeley, published at OSDI in 2022, and was subsequently adopted by mainstream inference engines such as vLLM, TensorRT-LLM, and TGI.
1
Sources
50%
Confidence
Long-term
Relevance
8/14/2026
First Seen
Sources
NVIDIA开源Switchyard：用Rust打造的AI任务调度引擎
Related Claims
VerifiedContinuous Batching (iteration-level scheduling) was first proposed by the Orca system, published in an OSDI 2022 paper.77% similarUnverifiedvLLM originated from a research team at UC Berkeley, and its core paper was published at SOSP (Symposium on Operating Systems Principles) in 2023.70% similarUnverifiedContinuous batching was systematically proposed in the 2023 PagedAttention paper from UC Berkeley.64% similarUnverifiedMemGPT由UC Berkeley于2023年提出，对Agent记忆架构进行系统性工程化探索63% similarUnverifiedvLLM was open-sourced by UC Berkeley in 2023 and increased LLM inference throughput by 2–4x through its PagedAttention technology.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/748159API
curl https://kongchang.com/api/v1/knowledge/claims/748159MCP
get_claim(id=748159)