2 code implementations • 7 Apr 2024 • Hongzheng Chen, Niansong Zhang, Shaojie Xiang, Zhichen Zeng, Mengjia Dai, Zhiru Zhang
For the GPT2 model, the inference latency of the Allo generated accelerator is 1. 7x faster than the NVIDIA A100 GPU with 5. 4x higher energy efficiency, demonstrating the capability of Allo to handle large-scale designs.
no code implementations • 23 Dec 2023 • Hongzheng Chen, Jiahao Zhang, Yixiao Du, Shaojie Xiang, Zichao Yue, Niansong Zhang, Yaohui Cai, Zhiru Zhang
Experimental results demonstrate our approach can achieve up to 13. 4x speedup when compared to previous FPGA-based accelerators for the BERT model.