Sudarshan Srinivasan
Papers
1
Total Citations
462
H-Index
1
About
Sudarshan Srinivasan is a leading researcher in computer architecture and hardware acceleration for deep learning, with a focus on efficient, high-performance systems for neural network training and inference. His most impactful work, "SIGMA: A Sparse and Irregular GEMM Accelerator with Flexible Interconnects for DNN Training" (2020), has garnered over 460 citations, establishing him as a key figure in designing specialized hardware for sparse and irregular matrix operations—a critical challenge in modern AI workloads. Srinivasan's contributions address the computational bottlenecks of general matrix-matrix multiplication (GEMM) in deep neural networks, proposing flexible interconnect fabrics that enable both dense and sparse computation without sacrificing efficiency. This work has influenced the development of next-generation accelerators for training large-scale models. Beyond SIGMA, his research spans energy-efficient architectures, memory systems, and domain-specific accelerators, with a track record of bridging algorithmic innovation with practical hardware design. Srinivasan's achievements are recognized through his publications in top venues like ISCA and HPCA, and his work continues to shape the trajectory of AI hardware, making him a pivotal figure for students and researchers interested in the intersection of computer architecture and machine learning.
Research Focus
Key Achievements
Top Papers
- 1