On S32K344 enabling prefetch and buffering on flash controller does not produce any visible difference when reading from flash memory. As far as I understand Technical Reference manual for S32Kxx enabling Prefetch and Data Buffers should result in shorter reading time, overall, from flash memory after first read.
However, that doesn't happen in my case - every read takes the same time regardless of Data Buffers or Prefetch being enabled or disabled.
My S32K344 also has disabled lockstep so we are using both Core1 and Core0. So for both PCRF0 and PCRF1 registers PxDP, PxCP, PxDB and PxCB fields are enabled. Reading is done from task that is run on Core0.

Following Technical Reference Manual registers are set before starting second core and code execution is in system RAM.

Is there anything else or any additional feature that needs to be activated for PFLASH to work as suggested above?
Reading from flash was tried on a couple of memory locations inside flash with different data varying between 50 bytes and 900 bytes.
Additionally, after inspecting memory location of PFCR registers in raw memory it seems that each core uses different endianness.
Core1 memory snapshot of PFCR0:

Core0 memory snapshot of PFCR0:

It is the same for PFCR1. It looks like from the likes of Technical Reference that this cannot be changed. Also, registers on other memory locations do not have this issue.
Another question is whether this is something that could impact work of PFLASH or it has nothing to do with the issue at hand.
Thanks in advance.