DOE OSTI · 3407878
TrioSim: A Lightweight Simulator for Large-Scale DNN Workloads on Multi-GPU Systems
Abstract
Deep Neural Networks (DNNs) have become increasingly capable of performing tasks ranging from image recognition to content generation. The training and inference of DNNs heavily rely on GPUs, as GPUs' massively parallel architecture delivers extremely high computing capability. With the growing complexity of DNNs and the size of training datasets, training DNNs with a large number of GPUs is becoming a prevalent strategy. Researchers have been exploring how to design software and hardware systems for GPU farms to achieve the best utilization, efficiency, and DNN accuracy during training or inference. However, when designing and deploying such systems, designers usually rely on testing on physical hardware platforms equipped with many GPUs, incurring high costs that are almost prohibitive for system designers to test different configurations and designs, even for highly resourceful companies. While an alternative solution is to test on GPU simulators, they are often too slow for these l
Keep this discovery
Explore connections, maps & timelines
Li, Ying [William & Mary, Williamsburg, VA, USA] (ORCID:0009000527370583), Bao, Yuhui [Northeastern University, Boston, MA, USA] (ORCID:0000000163083841), Wang, Gongyu [Lightmatter, Boston, MA, USA] (ORCID:0009000248007781), Mei, Xinxin [Jefferson Lab, Newport News, VA, USA] (ORCID:0000000310465269), Vaid, Pranav [Stanford University, Palo Alto, CA, USA] (ORCID:0009000972497574), Ghosh, Anandaroop [Lightmatter, Boston, MA, USA] (ORCID:0009000272947634), Jog, Adwait [University of Virginia, Charlottesville, VA, USA] (ORCID:0000000255257204), Bunandar, Darius [Lightmatter, Boston, MA, USA] (ORCID:0000000282185656), Joshi, Ajay [Lightmatter/Boston University, Boston, MA, USA] (ORCID:0000000232569942), Sun, Yifan [William & Mary, Williamsburg, VA, USA] (ORCID:0000000335326521). 2025-06-20. TrioSim: A Lightweight Simulator for Large-Scale DNN Workloads on Multi-GPU Systems. https://doi.org/10.1145/3695053.3731082
Cite the original work for its findings. Save a collection to share your selection of sources.