External Memory Interface Handbook Volume 3: Reference Material: For UniPHY-based Device Families

ID 683841
Date 3/06/2023
Public
Document Table of Contents

4.6.1. MPFE Operation Ordering

Operation ordering is defined and enforced within a port, but not between ports. All transactions received on a single port for overlapping addresses execute in order. Requests arriving at different ports have no guaranteed order of service, except when a first transaction has completed before the second arrives.

Avalon‑MM does not support write acknowledgement. When a port is configured to support Avalon-MM, you should read from the location that was previously written to ensure that the write operation has completed. When a port is configured to support AXI, the master accessing the port can safely issue a read operation to the same address as a write operation as soon as the write has been acknowledged. To keep write latency low, writes are acknowledged as soon as the transaction order is guaranteed—meaning that any operations received on any port to the same address as the write operation are executed after the write operation.

To reduce read latency, the single-port logic can return read data out of order to the multi-port logic. The returned data is rearranged to its initial order on a per port basis by the multi-port logic and no traffic reordering occurs between individual ports.

Read Data Handling

The MPFE contains a read buffer shared by all ports. If a port is capable of receiving returned data then the read buffer is bypassed. If the size of a read transaction is smaller than twice the memory interface width, the buffer RAM cannot be bypassed. The lowest memory latency occurs when the port is ready to receive data and the full width of the interface is utilized.