Tool/software: TI C/C++ Compiler
Hi,
in my code I have to copy some quite big structs, some of them are bigger than 256 words. For the smaller ones the expected combination of RPT and PREAD is generated and this performs well. For the bigger structs a call to memcpy (lcr memcpy) is generated which takes more than 10 times longer to execute.
I read in the compiler manual in the section about --rpt_threshold that under some conditions multiple RPTs may be generated, but whatever options I tried I never got this. Is there a way to force the compiler to generate as much RPTs it takes and avoid memcpy?
I can get it to work when I put multiple lines of memcpy in the code an copy the struct manuall in sections this way. But as my code is generated by Simulink this is not a good solution.
I use version 18.1.4.LTS.
Thanks,
Thomas.