Optimized Port Cache Coherent Implementation - 5.0 English

System Cache LogiCORE IP Product Guide (PG118)

Document ID
PG118
Release Date
2021-11-05
Version
5.0 English

When optimized port cache coherency is enabled, cached masters connected through the generic AXI4 slave ports are not included in the coherency domain. The reason for this is that the connection is pure AXI and not ACE, so it is not possible to snoop any master connected to a generic port.

All writes from a generic port remove corresponding line(s) from any MicroBlaze processor cache connected to an optimized port so that the new data is visible to the MicroBlaze processor. A read gets a snapshot of the current value of the coherency domain; if this value is stored locally (cached) in the AXI master it is the responsibility of that master to perform proper cache maintenance to remain coherent.

When cache coherency is enabled, exclusive transactions from the generic ports are disabled, and treated as normal transactions. The reason for this is that only the ACE transaction-based method with snoop messages is supported when cache coherency is enabled.