The communication performance of link-sharing method of buffer in NoC router: the relation between the communication performance and the number of banks
We have proposed a memory sharing method of the wormhole routed network-on-chip architecture. In our method, a memory is shared between multiple physical links by using the multi-port memory. In this paper, we present the pipeline processing method, and evaluate the communication performance in the...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Society for Science and Education (SSE), United Kingdom
2013
|
| Subjects: | |
| Online Access: | http://irep.iium.edu.my/33634/ http://irep.iium.edu.my/33634/ http://irep.iium.edu.my/33634/ http://irep.iium.edu.my/33634/2/20-72-1-PB.pdf |
| Summary: | We have proposed a memory sharing method of the wormhole routed network-on-chip architecture. In our method, a memory is shared between multiple physical links by using the multi-port memory. In this paper, we present the pipeline processing method, and evaluate the communication performance in the various situations. The pipeline of the proposed method has two courses of the route 1 and 2. The number of pipeline stages of route2 is 2 stages larger than the traditional router in order to use a shared memory. But delay is concealed if the capacity of a private buffer is enough. It is shown that the required number of memory banks required in multiport memory for 2-dimensional torus and 2-dimensional mesh networks is 8. Our proposed method yields high performance for both torus and mesh networks. Even this high performance is retained when the buffer size and the packet length are same. |
|---|