To distinguish between similar read
optimization features implemented across different FSS modules, the following
definitions are used:
- OSPI DAC Predicted Read
feature : This optimizes read performance by issuing a predicted read if
the current reads are complete. Please refer to OSPI Controller section for
more information.
- OSPI DAC Read PHY pipeline
mode : This feature is implemented by OSPI Controller to read extra
bytes based on Controller configuration at the end of a read from flash. This
ensures that chip select is not de-asserted during flash round-trip delay and
sample delay associated with PHY DLL. Please refer to section PHY Pipeline Mode for more information.
- FSS XIP Prefetcher :
The prefetcher automatically loads the next 32-byte block from flash
when specific conditions are met. This feature optimizes XIP performance in
two ways:
- Addresses CPU request patterns: CPU XIP requests typically occur at
random intervals. Requests are often non-consecutive
- Reduces latency: Proactively fetches the next cache line. Eliminates
waiting time for CPU requests. Particularly effective for linear
code execution
For detailed implementation and conditions, refer to the FSS XIP Prefetcher
- ECCM Read Pipeline Mode :
This feature assists in maintaining address continuity to flash by reading
MAC and ECCM words even if the request address is not in MAC or ECCM region. In
this case, the unused MAC or ECCM words are discarded. This is implemented by
the ECCM module since it is responsible for address translation and reserving
space in flash for ECCM and MAC. The purpose for this is to ensure OSPI can
receive continuous addresses such that it can skip sending command bytes to
flash. This feature does not need continuous requests. It will predict that the
next consecutive address will be requested and hence is optimizing the current
request to maintain address continuity. Please refer to section ECCM Read
Pipelined mode for more information.