Locality-Centric Data and Threadblock Management for Massive GPUs
Locality-Centric Data and Threadblock Management for Massive GPUs
复制标题
DOI：
10.1109/micro50266.2020.00086
复制
发表时间：
2020-10
期刊：
影响因子：
--
通讯作者：
Mahmoud Khairy;Vadim Nikiforov;D. Nellans;Timothy G. Rogers
中科院分区：
文献类型：
--
作者：
Mahmoud Khairy;Vadim Nikiforov;D. Nellans;Timothy G. Rogers
最近的工作表明，由于晶体管密度增长放缓、芯片成品率低以及光掩模的限制，在单个单片芯片中构建具有数百个SM的GPU将是不现实的。为了保持性能可伸缩性，建议将离散的GPU聚合到更大的虚拟GPU中，并将单个GPU分解为具有更大聚合芯片面积的多芯片模块。如果管理不当，这些方法会引入非一致内存访问(NUMA)效应，并导致性能和能效降低。为了克服这些影响，我们提出了一个完整的局部性感知数据管理(LADM)系统，旨在操作由多个离散设备组成的海量逻辑GPU，这些设备本身也由芯片组成。LADM有三个关键组件：以线程块为中心的索引分析、执行数据放置和线程块调度的运行时系统，以及自适应缓存插入策略。运行时将来自静态分析的信息与拓扑信息相结合，以主动优化数据放置、线程块调度和远程数据缓存，最大限度地减少芯片外流量。在未来的多GPU系统上，与现有的多GPU调度相比，LADM将片间存储流量减少了4倍，将系统性能提高了1.8倍。
Recent work has shown that building GPUs with hundreds of SMs in a single monolithic chip will not be practical due to slowing growth in transistor density, low chip yields, and photoreticle limitations. To maintain performance scalability, proposals exist to aggregate discrete GPUs into a larger virtual GPU and decompose a single GPU into multiple-chip-modules with increased aggregate die area. These approaches introduce non-uniform memory access (NUMA) effects and lead to decreased performance and energy-efficiency if not managed appropriately. To overcome these effects, we propose a holistic Locality-Aware Data Management (LADM) system designed to operate on massive logical GPUs composed of multiple discrete devices, which are themselves composed of chiplets. LADM has three key components: a threadblock-centric index analysis, a runtime system that performs data placement and threadblock scheduling, and an adaptive cache insertion policy. The runtime combines information from the static analysis with topology information to proactively optimize data placement, threadblock scheduling, and remote data caching, minimizing off-chip traffic. Compared to state-of-the-art multi-GPU scheduling, LADM reduces inter-chip memory traffic by 4× and improves system performance by 1.8× on a future multi-GPU system.