Buffer Pocketing and Pre-Checking on Buffer Utilization

Problem statement: In this study, the paper proposes the development of Effective Buffer Utilization on Adaptive Router with Buffer Pocketing and Pre-Buffer Checking Technique. Network-on-Chip could be prepared more capable by tricky faster routers. By using improved buffers, superior number of ports and channels, adaptive routing, all of which acquire key of overheads in hardware costs. Approach: This technique will improve communication efficiency without increasing the buffer size with support of input buffer space feedback controller in an input channel. A Buffer-Pocketing system enables the input channels to use the unused buffer from another channel at runtime that have not enough buffer space to utilize as per the input flits. The current buffer status in a router could be updated on each cycle of data flits transmitted to buffer space controller with the help of Router Monitor Sensor (RMS). Results: Implementation results of the proposed design for a 64-bit 4 input-buffer router show a reduction of the average packet transmission latency and an increase of the average transmission flits. The results show that the proposed design can reduce the cycles required for transmitting a fixed number of packets, when compared to that without buffer stealing. Conclusion: The study confirmed that the pre-buffer checking and feedback collecting from the router design take the place of the original design in terms of both throughput and latency. Thus, BS is more robust in handling hardware overhead ratio.

[1]  Doug Burger,et al.  Implementation and Evaluation of On-Chip Network Architectures , 2006, 2006 International Conference on Computer Design.

[2]  Li-Shiuan Peh,et al.  Leakage power modeling and optimization in interconnection networks , 2003, ISLPED '03.

[3]  Lei Gao,et al.  A dynamically-allocated virtual channel architecture with congestion awareness for on-chip routers , 2008, 2008 45th ACM/IEEE Design Automation Conference.

[4]  Wei Hwang,et al.  2-level FIFO architecture design for switch fabrics in network-on-chip , 2006, 2006 IEEE International Symposium on Circuits and Systems.

[5]  Timothy Mattson,et al.  A 48-Core IA-32 message-passing processor with DVFS in 45nm CMOS , 2010, 2010 IEEE International Solid-State Circuits Conference - (ISSCC).

[6]  Jin Liu,et al.  A DAMQ shared buffer scheme for network-on-chip , 2007 .

[7]  Masanori Hashimoto,et al.  Statistical analysis of clock skew variation in H-tree structure , 2005, Sixth international symposium on quality electronic design (isqed'05).

[8]  Sriram R. Vangal,et al.  A 5-GHz Mesh Interconnect for a Teraflops Processor , 2007, IEEE Micro.

[9]  Masanori Hashimoto,et al.  Statistical Analysis of Clock Skew Variation in H-Tree Structure , 2005, IEICE Trans. Fundam. Electron. Commun. Comput. Sci..

[10]  Stephen W. Keckler,et al.  Regional congestion awareness for load balance in networks-on-chip , 2008, 2008 IEEE 14th International Symposium on High Performance Computer Architecture.