Performance Evaluation of GraphCore IPU-M2000 Accelerator for Text Detection Application

作者：Nupur Sumeet, Karan Rawat, Manoj Nambiar · 年份：2022 · DOI：10.1145/3491204.3527469 · 被引用次数：5 · 研究领域：Parallel Computing and Optimization Techniques、Graph Theory and Algorithms、Ferroelectric and Negative Capacitance Devices

The large compute load and memory footprint of modern deep neural networks motivates the use of accelerators for high through- put deployments in application spanning multiple domains. In this paper, we evaluate throughput capabilities of a comparatively new hardware from Graphcore, IPU-M2000 that supports massive par- allelism and in-memory compute. For a text detection model, we measured the throughput and power variations with batch size. We also evaluate compressed versions of this model and analyze perfor- mance variation with model precision. Additionally, we compare IPU (Intelligence Processing Unit) results with state-of-the-art GPU and FPGA deployments of a compute intensive text region detec- tion application. Our experiments suggest, IPU supports superior throughput, 27×, 1.89×, and 1.56× as compared to CPU, FPGA DPU and A100 GPU, respectively for text detection application.