Embedded Design Handbook

ID 683689
Date 8/28/2023
Public
Document Table of Contents

7.1.1.2.2. Pipelining Algorithms

A common problem in systems with multiple Avalon® -MM master ports is competition for shared resources. You can improve performance by pipelining the algorithm and buffering the intermediate results in separate on-chip memories. The figure below illustrates this approach. Two hardware accelerators write their intermediate results to on-chip memory. The third module writes the final result to an off-chip memory. Storing intermediate results in on-chip memories reduces the I/O throughput required of the off-chip memory. By using on-chip memories as temporary storage you also reduce read latency because on-chip memory has a fixed, low-latency access time.

Figure 270. Using On-Chip Memory to Achieve High Performance

To learn more about optimizing memory design refer to the Memory System Design chapter of this handbook.