Gennady Pekhimenko
Associate Professor
Leads the EcoSystem group on hardware acceleration and systems for machine learning training.
Gennady Pekhimenko's lab at Toronto works on accelerators, chips & architecture and efficient training & inference. 28 papers since 2023, 19 at top venues, mostly ASPLOS and MLSys. 12 PhD students and 3 master's students. 25 alumni, most often to IBM and Amazon. Applications close .
At a glance
- papers, 2023–2026
- 7/ year
- at T1 venues
- 68%
- PhD students
- 12
- MS students
- 3
- alumni listed
- 25
Papers by yearfrom DBLP, 2023–2026
- 8 papers2026 · T1 5 · collecting
- 8 papers2025 · T1 7
- 5 papers2024 · T1 3
- 7 papers2023 · T1 4
T1 venuesOtherVenues: ASPLOS, MLSys, EuroSys, ICLR, ACL
Research focuspaper titles that use each topic's words
- Efficient training & inference11 papers
- Accelerators, chips & architecture1 paper
- ecosystem0 papers
- hardware acceleration0 papers
- learning training0 papers
Paper list
From DBLP.
20268 papers
- CVPR (T1)BHCast: Unlocking Black Hole Plasma Dynamics from a Single Blurry Image with Long-Term Forecasting (opens in a new tab)
- ISCA (T1)DCC: Data-Centric Compilation of Machine Learning Kernels for Processing-In-Memory Architectures (opens in a new tab)
- ICLR (T1)DPQuant: Efficient and Private Model Training via Dynamic Quantization Scheduling (opens in a new tab)
- IEEE Comput. Archit. Lett.Fusing Adds and Shifts for Efficient Dot Products (opens in a new tab)
- CGOHexcute: A Compiler Framework for Automating Layout Synthesis in GPU Programs (opens in a new tab)
- EuroSys (T1)LoRAFusion: Efficient LoRA Fine-Tuning for LLMs (opens in a new tab)
- CAISSwiftFusion: Scalable Sequence Parallelism for Distributed Inference of Diffusion Transformers on GPUs (opens in a new tab)
- ASPLOS (T1)Tilus: A Tile-Level GPGPU Programming Language for Low-Precision Computation (opens in a new tab)
20258 papers
- ACL (T1)APPL: A Prompt Programming Language for Harmonious Integration of Programs and Large Language Model Prompts (opens in a new tab)
- ASPLOS (T1)HybridTier: an Adaptive and Lightweight CXL-Memory Tiering System (opens in a new tab)
- EuroSys (T1)Mist: Efficient Distributed Training of Large Language Models via Memory-Parallelism Co-Optimization (opens in a new tab)
- SIGMETRICSPyGim: An Efficient Graph Neural Network Library for Real Processing-In-Memory Architectures (opens in a new tab)
- MLSys (T1)ScaleFusion: Scalable Inference of Spatial-Temporal Diffusion Transformers for High-Resolution Long Video Generation (opens in a new tab)
- MLSys (T1)Seesaw: High-throughput LLM Inference via Model Re-sharding (opens in a new tab)
- ASPLOS (T1)Tally: Non-Intrusive Performance Isolation for Concurrent Deep Learning Workloads (opens in a new tab)
- ICML (T1)TypyBench: Evaluating LLM Type Inference for Untyped Python Repositories (opens in a new tab)
20245 papers
- PACTBOOM: Use your Desktop to Accurately Predict the Performance of Large Deep Neural Networks (opens in a new tab)
- ICLR (T1)Guaranteed Approximation Bounds for Mixed-Precision Neural Operators (opens in a new tab)
- EuroSys (T1)Minuet: Accelerating 3D Sparse Convolutions on GPUs (opens in a new tab)
- MLSys (T1)Proteus: Preserving Model Confidentiality during Graph Optimizations (opens in a new tab)
- ICSSylva: Sparse Embedded Adapters via Hierarchical Approximate Second-Order Information (opens in a new tab)
20237 papers
- USENIX ATCArbitor: A Numerically Accurate Hardware Emulation Tool for DNN Accelerators (opens in a new tab)
- Nat. Mac. Intell.Federated benchmarking of medical artificial intelligence with MedPerf (opens in a new tab)
- MICRO (T1)Grape: Practical and Efficient Graphed Execution for Dynamic Deep Neural Networks on GPUs (opens in a new tab)
- ASPLOS (T1)Hidet: Task-Mapping Programming Paradigm for Deep Learning Tensor Programs (opens in a new tab)
- MLSys (T1)Hotline Profiler: Automatic Annotation and A Multi-Scale Timeline for Visualizing Time-Use in DNN Training (opens in a new tab)
- ASPLOS (T1)TiLT: A Time-Centric Approach for Stream Query Optimization and Parallelization (opens in a new tab)
- APLASTorchProbe: Fuzzing Dynamic Deep Learning Compilers (opens in a new tab)
Where alumni went25 listed, 64% to industry
Industry 7Academia 4
Most often to
- IBM
- Amazon
- ByteDance
Also to
- Carnegie Mellon University
- MIT
- UC BerkeleyGrad School EECS
Apply to TorontoFall 2027
Computer Science MSc/PhD
- Round
- Single round (Dec 1)
- Opens
- —
- GRE
- optional
- English
- TOEFL/IELTS if applicable
- Fee
- CAD 135
Confirmed on the program page.
Similar labsnearest by research summary
- CNTsinghua
Hongyang Jia
- Accelerators, chips & architecture
- USUCSD
Tajana Šimunić Rosing
- Accelerators, chips & architecture
- GBUCL
Adnan Mehonic
- Accelerators, chips & architecture
- GBImperial
Christos Bouganis
- Accelerators, chips & architecture
- USNYU
Brandon Reagen
- Accelerators, chips & architecture
- CNSJTU
Yu Feng
- Efficient training & inference
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- DBLP
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
Torontocloses