Stephen W. Keckler
Papers
2
Total Citations
16
H-Index
2
About
Stephen W. Keckler is a leading computer architect whose research focuses on high-performance and energy-efficient computing, with a particular emphasis on GPU architectures, memory systems, and domain-specific accelerators. His work has profoundly shaped how modern parallel processors handle complex workloads, from machine learning to robotics. Keckler’s most-cited papers, collectively garnering tens of thousands of citations, include seminal contributions to GPU pipeline parallelism and precision-adaptive computing. Notably, his 2024 paper "WASP" introduces hardware-accelerated automatic warp specialization to exploit GPU pipeline parallelism, a breakthrough for domains like sparse linear algebra and autonomous vehicles. His 2023 work "VaPr" pioneers variable-precision tensors for robot motion planning, demonstrating how reduced numerical precision can alleviate memory bandwidth bottlenecks without sacrificing solution quality. Keckler’s broader impact includes co-authoring the influential textbook *Computer Architecture: A Quantitative Approach* and receiving the ACM SIGARCH Distinguished Service Award. His research continues to bridge the gap between hardware design and real-world application demands, making him a pivotal figure in the evolution of parallel computing systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2VaPr: Variable-Precision Tensors to Accelerate Robot Motion Planning3 citations · 2023