P. (Saday) Sadayappan

About

P. (Saday) Sadayappan is a Professor in the School of Computing at the University of Utah. His research focuses on compiler optimization for high-performance computing, optimization of sparse/dense matrix and tensor computations, and algorithm-architecture co-design. He leads projects on tensor applications, sparse computations, and data locality optimization.

Selected publications

• FaSTCC: Fast Sparse Tensor Contractions on CPUs
• CoNST: Code Generator for Sparse Tensor Networks
• Accelerated Auto-Tuning of GPU Kernels for Tensor Computations
• Communication Optimization for Distributed Execution of Graph Neural Networks
• Effective Performance Modeling and Domain-Specific Compiler Optimization of CNNs for GPUs
• Comprehensive Accelerator-Dataflow Co-Design Optimization for Convolutional Neural Networks
• IOOpt: Automatic Derivation of I/O Complexity Bounds for Affine Programs
• Analytical Characterization and Design Space Exploration for Optimization of CNNs
• Brief Announcement: Efficient Distributed Algorithms for Convolutional Neural Networks
• Efficient Tiled Sparse Matrix Multiplication Through Matrix Signatures