论文标题
您只搜索一次：单级DNN/ACCELERATOR共同设计的快速自动化框架
You Only Search Once: A Fast Automation Framework for Single-Stage DNN/Accelerator Co-design
论文作者
论文摘要
DNN/ACCELERATOR共同设计在改善QOR和性能方面具有巨大的潜力。典型方法将设计流分为两个阶段：（1）设计具有高精度的应用程序特异性DNN模型； （2）构建一个考虑DNN特定特征的加速器。但是，在构建基于特定的基于神经网络的系统时，它可能无法承诺最高的综合分数和其他与硬件相关的约束（例如，延迟，能源效率）结合起来。在这项工作中，我们提出了一个单级自动化框架YOSO，旨在生成软件和硬件的最佳解决方案，该解决方案可以灵活地平衡准确性，功率和QoS的目标。与基线收缩期阵列加速器和CIFAR10数据集上的两阶段方法相比，我们分别在相同的精度上达到1.42x〜2.29x的能量或1.79 x〜3.07x的延迟降低，分别用于不同用户指定的能量和延迟优化约束。
DNN/Accelerator co-design has shown great potential in improving QoR and performance. Typical approaches separate the design flow into two-stage: (1) designing an application-specific DNN model with high accuracy; (2) building an accelerator considering the DNN specific characteristics. However, it may fail in promising the highest composite score which combines the goals of accuracy and other hardware-related constraints (e.g., latency, energy efficiency) when building a specific neural-network-based system. In this work, we present a single-stage automated framework, YOSO, aiming to generate the optimal solution of software-and-hardware that flexibly balances between the goal of accuracy, power, and QoS. Compared with the two-stage method on the baseline systolic array accelerator and Cifar10 dataset, we achieve 1.42x~2.29x energy or 1.79x~3.07x latency reduction at the same level of precision, for different user-specified energy and latency optimization constraints, respectively.
