We propose ZnG, a new GPU-SSD integrated architecture, which can maximize the memory capacity in a GPU and address performance penalties imposed by an SSD. Specifically, ZnG replaces all GPU internal DRAMs with an ultra-low-latency SSD to maximize the GPU memory capacity. ZnG further removes performance bottleneck of the SSD by replacing its flash channels with a high-throughput flash network and integrating SSD firmware in the GPU's MMU to reap the benefits of hardware accelerations. Although flash arrays within the SSD can deliver high accumulated bandwidth, only a small fraction of such bandwidth can be utilized by GPU's memory requests due to mismatches of their access granularity. To address this, ZnG employs a large L2 cache and flash registers to buffer the memory requests. Our evaluation results indicate that ZnG can achieve 7.5x higher performance than prior work.

中文翻译：

---

ZnG：使用新闪存构建 GPU 多处理器以实现可扩展的数据分析

我们提出了 ZnG，一种新的 GPU-SSD 集成架构，它可以最大化 GPU 的内存容量并解决 SSD 带来的性能损失。具体来说，ZnG 将所有 GPU 内部 DRAM 替换为超低延迟 SSD，以最大化 GPU 内存容量。ZnG 通过将其闪存通道替换为高吞吐量闪存网络并在 GPU 的 MMU 中集成 SSD 固件以获取硬件加速的好处，进一步消除了 SSD 的性能瓶颈。尽管 SSD 中的闪存阵列可以提供高累积带宽，但由于访问粒度的不匹配，GPU 的内存请求只能使用此类带宽的一小部分。为了解决这个问题，ZnG 采用了大型 L2 缓存和闪存寄存器来缓冲内存请求。我们的评估结果表明 ZnG 可以达到 7。