ID	原文	译文
52977	该架构采用基于时间片的调度方法，实现了多视觉任务的高效并行处理。	A time slice-based scheduling method is further proposed to conduct the multi-vision task in parallel and the network parameter sharing is utilized to achieve significant computational reuse.
52978	实验结果表明，与目前主流的移动图形处理器(GPU)相比，本文提出的加速器能够在多任务推理阶段减少65. 80%的乘法运算量，同时获得了 11倍的性能提升。	It is evaluated in experiments that the proposed accelerator removes 65. 80% of the repetitive operations from the multi-simultaneous inference tasks and achieves 11 × performance speedup over the state-of-the-art mobile graphics processing unit(GPU).
52979	以非易失性存储器(NVM)作为主存且以动态随机存取存储器(DRAM)作为片外高速缓存(EC)，是一种可以满足大数据应用内存容量需求的新型混合内存结构(ECNVM)。	To satisfy the memory capacity requirement of big-data applications, a promising architecture, in which the non-volatile memory(NVM) is used as main memory and dynamic random access memory(DRAM) is taken as external cache(EC), is proposed, referred to as EC-NVM.
52980	与ICDRAM相比，EC-NVM在容量比、延迟比方面均有显著不同，导致在IC-DRAM场景下的设计方法和优化策略直接迁移到EC-NVM上未必有良好的效果。	Compared to traditional memory architecture that involves the last level cache(LLC) and DRAM(referred to as IC-DRAM), EC-NVM is significantly different in terms of capacity and latency, thus traditional design cannot be directly borrowed without reconsideration.
52981	本文评测了 EC-NVM的体系结构特性(包括高速缓存粒度、关联度、替换算法、预取算法等)，获得了指导ECNVM结构的设计和优化的一系列发现。	This paper explores the characteristics and architectural implications of EC-NVM in diverse dimensions including cache line granularity, associativity, replacement policy and prefetching methods. All findings provide valuable hints for the designers of EC-NVM.
52982	随着深度神经网络对算力的需求不断增加，传统通用处理器在完成推理运算过程中出现了性能低、功耗高的缺点，	With the increasing demand for computing power of deep neural networks, traditional general-purpose processors have the disadvantages of low performance and high power consumption in the process of completing inference operations.
52983	因此通过专用硬件对深度神经网络进行加速逐步成为了深度神经网络的重要发展趋势。	Therefore, it has become an important development trend to accelerate deep neural networks through dedicated hardware.
52984	现场可编程门阵列(FPGA)具有重构性强、开发周期短以及性能优越等优点，	Field programmable gate array(FPGA) has the advantages of strong reconfigurability, short development cycle, and superior performance.
52985	适合用作深度神经网络的硬件加速平台。	It is very suitable as a hardware acceleration platform for deep neural networks.
52986	英伟达深度学习加速器(NVDLA)是英伟达开源的神经网络硬件加速器，	NVIDIA deep learning accelerator(NVDLA) is NVIDIA's open source neural network hardware accelerator.