Benchmarking GPU-Accelerated Edge Devices

摘要

We evaluate one of the state-of-the-art GPUaccelerated edge devices in this paper. We perform a set of deep learning benchmarks on the device to measure its performance. By comparing its performance with other GPUaccelerated systems in different computing platforms, we assess thecomputationalcapabilityofthemodernedgedevicesequipped with a signi?cant amount of hardware parallelism.