Torsten Hoefler
Full Professor
Researches high-performance computing and datacenter architecture, including scalable training systems for large machine learning models, networking and climate simulation.
Torsten Hoefler's lab at ETH Zurich works on distributed, cloud & edge systems, accelerators, chips & architecture and physics, climate & scientific computing. 100 papers since 2023, 16 at top venues, mostly ICLR and DAC. 18 PhD students and 2 master's students. 65 alumni, most often to IBM and Microsoft.
At a glance
- papers, 2023–2026
- 26/ year
- at T1 venues
- 16%
- PhD students
- 18
- MS students
- 2
- alumni listed
- 65
Papers by yearfrom the lab homepage, 2023–2026
- 17 papers2026 · T1 3 · collecting
- 26 papers2025 · T1 2
- 29 papers2024 · T1 7
- 28 papers2023 · T1 4
T1 venuesOtherVenues: ICLR, DAC, TPAMI, NeurIPS, AAAI
Research focuspaper titles that use each topic's words
- Distributed, cloud & edge systems15 papers
- Physics, climate & scientific computing9 papers
- Accelerators, chips & architecture3 papers
- datacenter3 papers
- training systems0 papers
Paper list
From the lab homepage.
202617 papers
- In Proceedings of the Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), presented in San Diego, California, United StatesApertus: Democratizing Open and Compliant LLMs for Global Language Environments (opens in a new tab)
- ICLR (T1)Beyond Outliers: A Study of Optimizers Under Quantization (opens in a new tab)
- IEEE Transactions on Parallel and Distributed Systems.Cost-Effective Empirical Performance Modeling (opens in a new tab)
- TPAMI (T1)Demystifying Higher-Order Graph Neural Networks (opens in a new tab)
- In Proceedings of the IEEE Interational Parallel and Distributed Processing Symposium (IPDPS'26), presented in New Orleans, LA, USAEvalNet: A Practical Toolchain for Generation and Analysis of Extreme-Scale Interconnects (opens in a new tab)
- IEEE/ACM Transactions on Networking. Jan.Flowcut Switching: High-Performance Adaptive Routing with In-Order Delivery Guarantees (opens in a new tab)
- IEEE Computer.In-Network Collective Operations: Game Changer or Challenge for AI Workloads? (opens in a new tab)
- DAC (T1)Network Design for Wafer-Scale Systems with Wafer-on-Wafer Hybrid Bonding (opens in a new tab)
- In ISC High Performance Research Paper Proceedings ( International Conference), presented in Hamburg, Germany, Jun.PICO: Performance Insights for Collective Operations (opens in a new tab)
- ICML. ETH Zurich. presented in Seoul, South Korea, Jul. , Just AcceptedProcess Reward Agents for Steering Knowledge-Intensive Reasoning (opens in a new tab)
- Geoscientific Model Development.psit 1.0: a system to compress Lagrangian flows (opens in a new tab)
- In Proceedings of the European Conference on Computer Systems (EuroSys’26), presented in Edinburgh, United Kingdom, Apr.REPS: Recycled Entropy Packet Spraying for Adaptive Load Balancing and Failure Mitigation (opens in a new tab)
202526 papers
- In CGO '25: Proceedings of the ACM/IEEE International Symposium on Code Generation and OptimizationA Priori Loop Nest Normalization: Automatic Loop Scheduling in Complex Applications (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov.Ab-initio Quantum Transport with the GW Approximation, 42,240 Atoms, and Sustained Exascale Performance (opens in a new tab)
- In Proceedings of the International Conference on Artificial Intelligence and Statistics (AISTATS'25), presented in Mai Khao, Thailand, MayAll models are wrong, some are useful: Model Selection with Limited Labels (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov.ATLAHS: An Application-centric Network Simulator Toolchain for AI, HPC, and Distributed Storage (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov.Bine Trees: Enhancing Collective Operations by Optimizing Communication Locality (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov.C.A.T.S.: Memory and Control Flow Tracing for Whole-Program Performance Analysis (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov.Computing the Full Earth System at 1km Resolution (opens in a new tab)
- In Proceedings of the IEEE International Symposium on Workload Characterization (IISWC), presented in Irvine, CA, USA, IEEE Press, Oct.Confidential LLM Inference: Performance and Cost Across CPU and GPU TEEs (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'25), presented in St. Louis, MO, USA, Nov.Core Hours and Carbon Credits: Incentivizing Sustainability in HPC (opens in a new tab)
- ACM Trans. Archit. Code Optim..Cppless: Productive and Performant Serverless Programming in C++ (opens in a new tab)
- In Proceedings of the USENIX Conference on Usenix Annual Technical Conference, presented in Boston, MA, USACrossPipe: Towards Optimal Pipeline Schedules for Cross-Datacenter Training (opens in a new tab)
- In Proceedings of the IEEE International Conference on Cluster Computing (CLUSTER), presented in Edinburgh, United KingdomDaCe AD: Unifying High-Performance Automatic Differentiation for Machine Learning and Scientific Computing (opens in a new tab)
202429 papers
- In USENIX Symposium on Networked Systems Design and Implementation (NSDI '24), presented in Santa Clara, CA, USAA High-Performance Design, Implementation, Deployment, and Evaluation of the Slim Fly Network (opens in a new tab)
- IEEE Transactions on Computers.A High-Performance, Energy-Efficient Modular DMA Engine Architecture (opens in a new tab)
- In Proceedings of the ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming (PPoPP'24), presented in Edinburgh, United KingdomArrow Matrix Decomposition: A Novel Approach for Communication-Efficient Sparse Matrix Multiplication (opens in a new tab)
- IEEE Transactions on Parallel and Distributed Systems.AutoDDL: Automatic Distributed Deep Learning With Near-Optimal Bandwidth Cost (opens in a new tab)
- VLDB (T1)Chameleon: A Heterogeneous and Disaggregated Accelerator System for Retrieval-Augmented Language Models (opens in a new tab)
- ICML (T1)DiffDA: a Diffusion model for weather-scale Data Assimilation (opens in a new tab)
- Nature Computational Science.Digital Twins of Earth and the Computing Challenge of Human Interaction (opens in a new tab)
- Earth System Science Data (ESSD).Earth Virtualization Engines (EVE) (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'24), presented in Atlanta, GA, USAExploring GPU-to-GPU Communication: Insights into Supercomputer Interconnects (opens in a new tab)
- In Proceedings of the International Symposium on High-Performance Parallel and Distributed Computing, presented in Pisa, Italy, ACM, ISBN: 9798400704130, Jun.FaaSKeeper: Learning from Building Serverless Services with ZooKeeper as an Example (opens in a new tab)
- AAAI (T1)Graph of Thoughts: Solving Elaborate Problems with Large Language Models (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'24), presented in Atlanta, GA, USAHigh Performance Unstructured SpMM Computation Using Tensor Cores (opens in a new tab)
202328 papers
- CGOBridging Control-Centric and Data-Centric Optimization (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'23), presented in Denver, CO, USA, Association for Computing Machinery, ISBN: 979-8-400701-09-2, Nov.Co-design Hardware and Algorithm for Vector Search (opens in a new tab)
- ICLR (T1)Compressing multidimensional weather and climate data into neural networks (opens in a new tab)
- IEEE Computer.Data Center Ethernet and Remote Direct Memory Access: Issues at Hyperscale (opens in a new tab)
- Nature Reviews Earth and Environment.Deep learning and a changing economy in weather and climate prediction (opens in a new tab)
- ACM Comput. Surv..Demystifying Graph Databases: Analysis and Taxonomy of Data Organization, System Designs, and Graph Queries (opens in a new tab)
- IEEE/CVF International Conference on Computer Vision (ICCV). Oct.Differentiable Transportation Pruning (opens in a new tab)
- Vol 66, Nr. 5, In Communications of the ACMDisentangling hype from practicality: On realistically achieving quantum advantage (opens in a new tab)
- Computing in Science and Engineering (CiSE).Earth Virtualization Engines -- A Technical Perspective (opens in a new tab)
- Jun.FMI: Fast and Cheap Message Passing for Serverless Functions (opens in a new tab)
- In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'23), Association for Computing Machinery, ISBN: 979-8-400701-09-2, Nov.FuzzyFlow: Leveraging Dataflow To Find and Squash Program Optimization Bugs (opens in a new tab)
- ICLR (T1)GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers (opens in a new tab)
Where alumni went65 listed, 50% to industry
Industry 2Academia 2
Most often to
- IBM
- Microsoft
Also to
- ETH Zurich
Apply to ETH ZurichRolling
Not announced yet
Computer Science Doctorate (D-INFK)
- Round
- Rolling (no deadline)
- Opens
- —
- GRE
- See program page
- English
- Proficiency in written and spoken English
- Fee
- See program page
Confirmed on the program page.
Similar labsnearest by research summary
- USNYU
Sai Qian Zhang
- Accelerators, chips & architecture
- CHETH Zurich
Luca Benini
- Accelerators, chips & architecture
- CNSJTU
Xiaofeng Hou
- Accelerators, chips & architecture
- GBCambridge
Robert Mullins
- Accelerators, chips & architecture
- CNPeking
Yufei Ma
- Accelerators, chips & architecture
- CNSJTU
Jieru Zhao
22 T1
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.