Mingxing Zhang
Associate Professor
Works on parallel and distributed systems, and initiated the open-source KVCache.AI projects Mooncake and KTransformers for large-language-model serving.
Mingxing Zhang's lab at Tsinghua works on distributed, cloud & edge systems and efficient training & inference. 40 papers since 2023, 30 at top venues, mostly SOSP and ASPLOS. 33 PhD students and 4 master's students. 55 alumni, most often to Huawei and Tencent. Applications close .
At a glance
- papers, 2023–2026
- 10/ year
- at T1 venues
- 75%
- PhD students
- 33
- MS students
- 4
- alumni listed
- 55
Papers by yearfrom the lab homepage, 2023–2026
- 8 papers2026 · T1 7 · collecting
- 16 papers2025 · T1 12
- 8 papers2024 · T1 6
- 8 papers2023 · T1 5
T1 venuesOtherVenues: SOSP, ASPLOS, SIGMOD, EuroSys, ICDE
Research focuspaper titles that use each topic's words
- Efficient training & inference7 papers
- Distributed, cloud & edge systems5 papers
- serving3 papers
- open-source0 papers
- distributed systems0 papers
Paper list
From the lab homepage.
20268 papers
- SIGMOD (T1)Accelerating Stream Processing Engines via Hardware Offloading
- SIGMOD (T1)CMANNS: GPU-Accelerated Graph Index Construction for ANNS via Compute–Memory Disaggregation
- KDD (T1)Faico: Faithful and Complete Knowledge Graph Augmented Reasoning
- SIGMOD (T1)From Prefix Cache to Fusion RAG Cache: Accelerating LLM Inference in Retrieval-Augmented
- DAC (T1)GWrite: Offloading Replication in Disaggregated Memory Systems
- OSDI (T1)Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning
- ATCSplineFS: Adaptive Learned Directory Indexing for Memory-Constrained Filesystems
- DAC (T1)SSD Reforge: Dynamic Parity Management to Extend SSD Lifetime under Aging
202516 papers
- EuroSys (T1)AlloyStack: A Library Operating System for Serverless Workflow Applications
- ATCDSA-2LM: A CPU-Free Tiered Memory Architecture with Intel DSA
- OSDI (T1)Fork in the Road: Reflections and Optimizations for Cold Start Latency in Production Serverless Systems
- SOSP (T1)KTransformers: Unleashing the Full Potential of CPU/GPU Hybrid Inference for MoE Models
- SOSP (T1)Mantle: Efficient Hierarchical Metadata Management for Cloud Object Storage Services
- NeurIPS (T1)MigGPT: Harnessing Large Language Models for Automated Migration of Out-of-Tree Linux Kernel Patches Across Versions
- NeurIPS (T1)MoBA: Mixture of Block Attention for Long-Context LLMs
- ToSMooncake: A KVCache-centric Disaggregated Architecture for LLM Serving
- FASTMOONCAKE: Trading More Storage for Less Computation —— A KVCache-centric Architecture for Serving LLM Chatbot
- ICDE (T1)OOCC: One-round Optimistic Concurrency Control for Read-Only Disaggregated Transactions
- SIGCOMM (T1)Pegasus: A Universal Framework for Scalable Deep Learning Inference on the Dataplane
- ICDE (T1)Scaling Asynchronous Graph Query Processing via Partitioned Stateful Traversal Machines
20248 papers
- ICDE (T1)Enabling Efficient NVM-Based Text Analytics without Decompression
- ATCHydraRPC: RPC in the CXL Era
- VLDB (T1)Improving Graph Compression for Efficient Resource-Constrained Graph Analytics
- EuroSys (T1)Noctua: Towards Automated and Practical Fine-grained Consistency Analysis
- ASPLOS (T1)Scaling Up Memory Disaggregated Applications With Smart
- SOSP (T1)Skyloft: A General High-Efficient Scheduling Framework in User Space
- SOSP (T1)TrEnv: Transparently Share Serverless Execution Environments Across Different Functions and Nodes
- HPCCxCache: An Adaptive Compression Strategy for Metadata Cache
20238 papers
- ASPLOS (T1)Achieving Sub-second Pairwise Query over Evolving Graphs
- ATC'23Explore Data Placement Algorithm for Balanced Recovery Load Distribution
- SOSP (T1)Falcon: Fast OLTP Engine for Persistent Cache and Non-Volatile Memory
- ISLPEDMulti-objective optimization for Floating Point Mix-Precision Tuning
- ASPLOS (T1)NosWalker: A Decoupled Architecture for Out-of-Core Random Walk Processing
- SOSP (T1)Partial Failure Resilient Memory Management System for (CXL-based) Distributed Shared Memory
- SPAA ‘23PIM-trie: A Skew-resistant Trie for Processing-in-Memory.
- EuroSys (T1)TEA: A General-Purpose Temporal Graph Random Walk Engine
Where alumni went55 listed, 79% to industry
Industry 33Academia 9
Most often to
- Huawei
- Tencent
- Beijing HaiZhi XingTu Technology Co.
Also to
- Qiyuan Laboratory
- Institute of Software Chinese Academy of Sciences
- NYU
Apply to TsinghuaFall 2027
Computer Science PhD (international applicants)
- Round
- Main
- Opens
- GRE
- See program page
- English
- Per school catalog
- Fee
- RMB 800
Not yet confirmed for this intake.
Similar labsnearest by research summary
- CNTsinghua
Wenlai Zhao
- Distributed, cloud & edge systems
- HKCUHK
He Chen
- Distributed, cloud & edge systems
- CNSJTU
Qinya Li
- Distributed, cloud & edge systems
- HKCUHK
Zhenyu Yan
- Distributed, cloud & edge systems
- CNSJTU
Chengzhang Li
- Distributed, cloud & edge systems
- CNPeking
Xuanzhe Liu
- Distributed, cloud & edge systems
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- This site is shared with other labs, so counts may overlap.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
This lab shares a website with other labs. Member and alumni counts may include them.
Tiers are venue classifications, not lab rankings.
Tsinghuacloses