Leaving One Slot Empty: Flit Bubble Flow Control for Torus Cache-Coherent NoCs
Leaving One Slot Empty: Flit Bubble Flow Control for Torus Cache-Coherent NoCs
复制标题
DOI：
10.1109/tc.2013.2295523
复制
发表时间：
2015-03
影响因子：
3.7
通讯作者：
Sheng Ma;Zhiying Wang;Zonglin Liu;Natalie D. Enright Jerger
中科院分区：
文献类型：
--
作者：
Sheng Ma;Zhiying Wang;Zonglin Liu;Natalie D. Enright Jerger
短分组和长分组共存于高速缓存一致的NoC中。现有的环面网络设计不能有效地处理可变大小的数据包。对于无死锁操作，设计使用两个VC，这会对路由器频率产生负面影响。一些优化使用一个VC。然而，他们认为所有的数据包作为最大长度的数据包，低效地利用宝贵的缓冲区。我们提出了微片气泡流控制（FBFC），它保持一个自由的微片大小的缓冲槽，以避免死锁。FBFC使用一个VC，并且不将短数据包视为长数据包。它实现了高频率和高效的缓冲区利用率。FBFC在4 × 4环面中的合成流量方面比LBS和CBS分别高出92.8%和34.2%。在更大的网络中，增益增加;在8 × 8圆环中，增益分别为107.2%和40.1%。对于PARSEC工作负载，FBFC实现了比LBS平均13.0%的加速。我们的研究结果还表明，FBFC比LBS和CBS更省电，并且具有FBFC的环面比网格更省电。
Short and long packets co-exist in cache-coherent NoCs. Existing designs for torus networks do not efficiently handle variable-size packets. For deadlock free operations, a design uses two VCs, which negatively affects the router frequency. Some optimizations use one VC. Yet, they regard all packets as maximum-length packets, inefficiently utilizing the precious buffers. We propose flit bubble flow control (FBFC), which maintains one free flit-size buffer slot to avoid deadlock. FBFC uses one VC, and does not treat short packets as long ones. It achieves both high frequency and efficient buffer utilization. FBFC performs 92.8 and 34.2 percent better than LBS and CBS for synthetic traffic in a 4 × 4 torus. The gains increase in larger networks; they are 107.2 and 40.1 percent in an 8 × 8 torus. FBFC achieves an average 13.0 percent speedup over LBS for PARSEC workloads. Our results also show that FBFC is more power efficient than LBS and CBS, and a torus with FBFC is more power efficient than a mesh.