Information

• The core requests four sequential longwords in back-to-back requests, meaning there
are no core cycle delays except for stalls waiting for flash memory data to be
returned.
• None of the data is already stored in the cache or speculation buffer.
In this scenario, the sequence of events for accessing the four longwords is as follows:
1. The first longword read requires 4 to 7 core clocks. See Wait states for more
information.
2. Due to the 64-bit data bus of the flash memory, the second longword read takes only
1 core clock because the data is already available inside the FMC. While the data for
the second longword is being returned to the core, the FMC also starts reading the
third and fourth longwords from the flash memory.
3. Accessing the third longword requires 3 core clock cycles. The flash memory read
itself takes 4 clocks, but the first clock overlaps with the second longword read.
4. Reading the fourth longword, like the second longword, takes only 1 clock due to the
64-bit flash memory data bus.
27.6 Initialization and application information
The FMC does not require user initialization. Flash acceleration features are enabled by
default.
The FMC has no visibility into flash memory erase and program cycles because the Flash
Memory module manages them directly. As a result, if an application is executing flash
memory commands, the FMC's cache might need to be disabled and/or flushed to prevent
the possibility of returning stale data. Use the PFB0CR[CINV_WAY] field to invalidate
the cache in this manner.
Initialization and application information
K20 Sub-Family Reference Manual, Rev. 1.1, Dec 2012
580
Preliminary
Freescale Semiconductor, Inc.
General Business Information