Matrix Engines for High Performance Computing:A Paragon of Performance or Grasping at Straws?
arXiv:2010.14373
Abstract
Matrix engines or units, in different forms and affinities, are becoming a reality in modern processors; CPUs and otherwise. The current and dominant algorithmic approach to Deep Learning merits the commercial investments in these units, and deduced from the No.1 benchmark in supercomputing, namely High Performance Linpack, one would expect an awakened enthusiasm by the HPC community, too. Hence, our goal is to identify the practical added benefits for HPC and machine learning applications by having access to matrix engines. For this purpose, we perform an in-depth survey of software stacks, proxy applications and benchmarks, and historical batch job records. We provide a cost-benefit analysis of matrix engines, both asymptotically and in conjunction with state-of-the-art processors. While our empirical data will temper the enthusiasm, we also outline opportunities to misuse these dense matrix-multiplication engines if they come for free.
arXiv commentIEEE International Parallel and Distributed Processing Symposium 2021 (IPDPS'21)
References in corpus (10)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
- Rethinking Atrous Convolution for Semantic Image Segmentation
- Neural Collaborative Filtering
- A Survey on Coarse-Grained Reconfigurable Architectures from a Performance Perspective
- Accelerating Sparse Matrix-Matrix Multiplication with GPU Tensor Cores
- A Survey of FPGA Based Deep Learning Accelerators: Challenges and Opportunities
- A Survey of Numerical Methods Utilizing Mixed Precision Arithmetic
- Automatic Kernel Generation for Volta Tensor Cores