Buddy Compression：为GPU上的深度学习和HPC工作负载启用更大的内存
Buddy Compression: Enabling Larger Memory for Deep Learning and HPC Workloads on GPUs
摘要：
GPU提供的内存带宽数量级高于传统的CPU系统。然而，GPU设备内存往往相对较小，用户无法增加内存容量。本文介绍了Buddy Compression，这是一种增加GPU有效内存容量和带宽的方案，同时避免了传统内存扩展策略的缺点。Buddy Compression压缩GPU内存，将每个压缩内存条目在高速设备内存和较慢但较大的分解内存池（或系统内存）之间拆分。因此，高度可压缩的存储器条目可以完全从设备存储器访问，而不可压缩的条目使用设备上和设备外访问来获取其数据。增加有效GPU内存容量使我们能够运行更大内存占用量的HPC工作负载，以及比当前内存容量所允许的更大批量或模型的DL工作负载。我们表明，我们的解决方案在HPC工作负载上实现了2.2倍的平均压缩比，在DL工作负载上实现了1.5倍的平均压缩比，速度仅为1~ 2%。
英文摘要：
GPUs offer orders-of-magnitude higher memory bandwidth than traditional CPU-only systems. However, GPU device memory tends to be relatively small and the memory capacity can not be increased by the user. This paper describes Buddy Compression, a scheme to increase both the effective GPU memory capacity and bandwidth while avoiding the downsides of conventional memory-expanding strategies. Buddy Compression compresses GPU memory, splitting each compressed memory entry between high-speed device memory and a slower-but-larger disaggregated memory pool (or system memory). Highly-compressible memory entries can thus be accessed completely from device memory, while incompressible entries source their data using both on and off-device accesses. Increasing the effective GPU memory capacity enables us to run larger-memory-footprint HPC workloads and larger batch-sizes or models for DL workloads than current memory capacities would allow. We show that our solution achieves an average compression ratio of 2.2x on HPC workloads and 1.5x on DL workloads, with a slowdown of just 1~2%.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
