AnsweredAssumed Answered

How do you run an SDK example entirely from SDRAM?

Question asked by Andrew Hollom on Jan 29, 2020
Latest reply on Feb 3, 2020 by Andrew Hollom
Hello,
I am evaluating the Cortex M7 using the iMXRT1050EVKB and the IAR IDE with an IAR I-Jet. I have successfully managed to build some of the SDK demo projects like led_blinkey, and have incorporated some benchmarking code to validate its operation. So far, I have executed my benchmarks from FlexRAM (text in ITCM and data in DTCM) with processor caches on and off and noted no performance change (expected, as FlexRAM bypasses the caches), and I have also executed it with the text in ITCM and data in SEMC0 (SDRAM) and noted that performance degrades if the data cache is disabled but does not change if the instruction cache is disabled (expected, as SEMC0 uses the data cache and ITCM bypasses the instruction cache).
Now, what I'd really like to do is to put the text segment into SEMC0 (SDRAM) to measure performance relative to having it in ITCM. I have tried this and while the program downloads and pauses at main(), as soon as it is executed I get a MemManage fault.
I have attached the ICF file, so hopefully someone can suggest what requires altering.
Thanks in advance.

Outcomes