Fangxin Liu
Assistant Professor
Works on computer architecture and design automation, including LLM acceleration and AI compilation.
Fangxin Liu's lab at SJTU works on accelerators, chips & architecture and efficient training & inference. 82 papers since 2023. 7 PhD students and 6 master's students. Applications close .
At a glance
- papers, 2023–2026
- 21/ year
- at T1 venues
- 0%
- PhD students
- 7
- MS students
- 6
- alumni listed
- —
Papers by yearfrom the lab homepage, 2023–2026
- 33 papers2026 · T1 0 · collecting
- 24 papers2025 · T1 0
- 22 papers2024 · T1 0
- 3 papers2023 · T1 0
T1 venuesOther
Research focuspaper titles that use each topic's words
- Efficient training & inference38 papers
- Accelerators, chips & architecture7 papers
- compilation3 papers
- design automation0 papers
- computer architecture0 papers
Paper list
From the lab homepage.
202633 papers
- PPoPP '26Accelerating Sparse Transformer Inference on GPU
- ASPLOS '26ASDR: Exploiting Adaptive Sampling and Data Reuse for CIM-based Instant Neural Rendering
- PPoPP '26BEEMS: Boosting Machine Vision Efficiency via Computation Graph-Based Memory Smoothing
- ASPDAC '26BLADE: Boosting LLM Decoding's Communication Efficiency in DRAM-based PIM
- DAC '26Brain-Like Hyper-Dimensional Graph Learning System with Hardware-Efficient Adaptive Sparsity
- ACL '26Calibrated Speculative Decoding: Frequency-Guided Candidate Selection for Efficient Inference
- ISCA '26COSM: A Cooperative Scheduling Framework for Concurrent PIM and CPU Execution on Mobile Devices
- MICRO '26DeGS: A Scalable 3DGS Architecture via Decoupled Workload Parsing and Reorganization
- EMNLP '26DELTA: Decoupling Latent Heterogeneity in Asymmetric Low-Rank Compression
- MICRO '26Deltoris: Enabling Real-time VLA Inference in Embodied AI via Bit-level Sparsity and Speculative Inference
- ACM MM '26DNA-ViT: Developmental Neural Archiving for Storage-Efficient Vision Transformers
- ASPLOS '26EARTH: An Efficient MoE Accelerator with Entropy-Aware Speculative Prefetch and Result Reuse
202524 papers
- TCAD '25A Sub-10 μs In-Memory-Search Collision Detection Accelerator Based on RRAM-TCAMs
- DAC '25ALLMod: Exploring Area-Efficiency of LUT-based Large Number Modular Reduction via Hybrid Workloads
- ACM MM '25ASTER: Adaptive Dynamic Layer-Skipping for Efficient Transformer Inference via MDP
- TACO '25Attack and Defense: Enhancing Robustness of Binary Hyper-Dimensional Computing
- DAC '25BLOOM: Bit-Slice Framework for DNN Acceleration with Mixed-Precision
- HPCA '25CROSS: Compiler-Driven Optimization of Sparse DNNs Using Sparse/Dense Computation Kernels
- ICCAD '25DevTrace: Efficient and Fine-grained PCIe Transaction Tracing for Edge Intelligence Workloads
- DATE '25EVASION: Efficient KV Cache Compression via Product Quantization
- ASPDAC '25Exploiting Differential-Based Data Encoding for Enhanced Query Efficiency
- ISCA '25FATE: Boosting HDC Intelligence Performance with Flexible Numerical Data Type
- EMNLP '25FlexQuant: A Flexible and Efficient Dynamic Precision Switching Framework for LLM Quantization
- HPCA '25GSArch: Breaking Memory Barriers in 3D Gaussian Splatting Training via Architectural Support
202422 papers
- ASPLOS '24CMC: Video Transformer Acceleration via CODEC Assisted Matrix Condensing
- MICRO '24COMPASS: SRAM-Based Computing-in-Memory SNN Accelerator with Adaptive Spike Speculation
- DAC '24EOS: An Energy-Oriented Attack Framework for Spiking Neural Networks
- TPDS '24Exploiting Temporal-Unrolled Parallelism for Energy-Efficient SNN Acceleration
- ICCD '24HOLES: Boosting LLMs Efficiency with Hardware-friendly Lossless Encoding
- DAC '24HyperAttack: An Efficient Attack Framework for HyperDimensional Computing
- ASPDAC '24HyperFeel: An Efficient Federated Learning Framework Using Hyperdimensional Computing
- DAC '24INSPIRE: Accelerating Deep Neural Networks via Hardware-friendly Index-Pair Encoding
- ISLPED '24LowPASS: A Low power PIM-based accelerator with Speculative Scheme for SNNs
- ICCD '24Ninja: A Hardware Assisted System for Accelerating Nested Address Translation
- ASPDAC '24PAAP-HD: PIM-Assisted Approximation for Efficient Hyper-Dimensional Computing
- ICCD '24PS4: A Low Power SNN Accelerator with Spike Speculative Scheme
20233 papers
- IEEE TC '23ERA-BS: Efficiency of ReRAM-based PIM Accelerator with Fine-Grained Bit-Level Sparsity
- ICCAD '23HyperNode: An Efficient Node Classification Framework Using HyperDimensional Computing
- ICCD '23PSQ: An Automatic Search Framework for Data-Free Quantization on PIM-based Architecture
Where alumni went
Alumni are not listed on the lab homepage.
Apply to SJTUFall 2027
Computer Science PhD (international applicants)
- Round
- Self-funded
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- CNPeking
Bonan Yan
- Accelerators, chips & architecture
- USStanford
Kunle Olukotun
- Accelerators, chips & architecture
- GBUCL
Adnan Mehonic
- Accelerators, chips & architecture
- GBImperial
Christos Bouganis
- Accelerators, chips & architecture
- CNTsinghua
Xing Lin
- Accelerators, chips & architecture
- USUW
Ang Li
- Accelerators, chips & architecture
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- The alumni list could not be found on the lab site.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
SJTUcloses