Ilias Iliadis
International Journal On Advances In Networks And Services
The computational requirements of artificial intel- ligence workloads are growing exponentially. At the same time, Dennard scaling has ended and Moore’s law is winding down. In addition, more and more compute is moved towards the edge due to latency or localization constraints. These trends cre- ated an opportunity for specialized accelerators including field- programmable gate arrays (FPGAs), but the poor support and usability of today’s tools prevents FPGAs from being deployed at scale for deep neural network (DNN) inference applications. In this work, we propose an organic compiler — DOSA — that drastically lowers the barrier for deploying FPGAs. DOSA builds on the operation set architecture concept and mixes the DNN accelerator components generated by existing DNN-to-FPGA frameworks to produce an overall efficient solution. DOSA starts from DNNs represented in the community standard ONNX and automatically implements model- and data-parallelism, based on the performance targets and resource footprints provided by the user. Deploying a DNN using DOSA on 9 FPGAs exhibits a speedup of up to 52 times compared to a CPU and 18 times compared to a GPU.
Ilias Iliadis
International Journal On Advances In Networks And Services
Gloire Rubambiza, Braulio Dumba, et al.
EDGE 2023
Juan Miguel De Haro, Rubén Cano, et al.
IPDPS 2022
Haoran Qiu, Weichao Mao, et al.
ASPLOS 2024