Musfiqur Rahman Sanim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Musfiqur Rahman Sanim has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Gagan Agrawal
University of Georgia, Athens, GA, USA
2 shared publications - Miao Yin
University of Texas at Arlington, Arlington, TX, USA
2 shared publications - Wei Niu
University of Georgia, Athens, GA, USA
2 shared publications - Zhihao Shu
University of Georgia, Athens, GA, USA
2 shared publications - Bin Ren
College of William & Mary, Williamsburg, USA
1 shared publication - Hangyu Zheng
University of Georgia, Athens, GA, USA
1 shared publication
Latest Publications
- Open Access
- 22 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 783-797
FlashMem: Supporting Modern DNN Workloads on Mobile with GPU Memory Hierarchy Optimizations
- 22 March 2026
- Published by Association for Computing Machinery (ACM)
- p. 783-797
Open AccessFlashMem: Supporting Modern DNN Workloads on Mobile with GPU Memory Hierarchy Optimizations
- Open Access
- 27 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 916-931
SmartMem: Layout Transformation Elimination and Adaptation for Efficient DNN Execution on Mobile
This work is motivated by recent developments in Deep Neural Networks, particularly the Transformer architectures underlying applications such as ChatGPT, and the need for performing inference on mobile devices. Focusing on emerging transformers (specifically the ones with computationally efficient Swin-like architectures) and large models (e.g., Stable Diffusion and LLMs) based on transformers, we observe that layout transformations between the computational operators cause a significant slowdown in these applications. This paper presents SmartMem, a comprehensive framework for eliminating most layout transformations, with the idea that multiple operators can use the same tensor layout through careful choice of layout and implementation of operations. Our approach is based on classifying the operators into four groups, and considering combinations of producer-consumer edges between the operators. We develop a set of methods for searching such layouts. Another component of our work is developing efficient memory layouts for 2.5 dimensional memory commonly seen in mobile devices. Our experimental results show that SmartMem outperforms 5 state-of-the-art DNN execution frameworks on mobile devices across 18 varied neural networks, including CNNs, Transformers with both local and global attention, as well as LLMs. In particular, compared to DNNFusion, SmartMem achieves an average speedup of 2.8×, and outperforms TVM and MNN with speedups of 6.9× and 7.9×, respectively, on average.
- 27 April 2024
- Published by Association for Computing Machinery (ACM)
- p. 916-931
Open AccessSmartMem: Layout Transformation Elimination and Adaptation for Efficient DNN Execution on Mobile
This work is motivated by recent developments in Deep Neural Networks, particularly the Transformer architectures underlying applications such as ChatGPT, and the need for performing inference on mobile devices. Focusing on emerging transformers (specifically the ones with computationally efficient Swin-like architectures) and large models (e.g., Stable Diffusion and LLMs) based on transformers, we observe that layout transformations between the computational operators cause a significant slowdown in these applications. This paper presents SmartMem, a comprehensive framework for eliminating most layout transformations, with the idea that multiple operators can use the same tensor layout through careful choice of layout and implementation of operations. Our approach is based on classifying the operators into four groups, and considering combinations of producer-consumer edges between the operators. We develop a set of methods for searching such layouts. Another component of our work is developing efficient memory layouts for 2.5 dimensional memory commonly seen in mobile devices. Our experimental results show that SmartMem outperforms 5 state-of-the-art DNN execution frameworks on mobile devices across 18 varied neural networks, including CNNs, Transformers with both local and global attention, as well as LLMs. In particular, compared to DNNFusion, SmartMem achieves an average speedup of 2.8×, and outperforms TVM and MNN with speedups of 6.9× and 7.9×, respectively, on average.