My program project originally developed in 28335 + CCS3.3 + Code Generation V6.20, because the program involved a large number of matrix calculation, compilation is very slow, so I extracted the asm file generated by matrix calculation source Code, and created a new static Library project in CCS3.3. The .asm file was added to generate the static library, the static library size is 297kb, so the entire program compiled much faster.
But as the processing performance demand improve, I needed to port the original Code to 28377 + CCS6.1 + Code Generation V21.6.0. LTS. In order to improve compilation efficiency, I still extracte the .asm file generated by matrix calculation source Code, In order to generate. ASM File, the program project properties set-k –al –as refer to CCS3.3 program project settings, then create a new static Library project in CCS6.1. The asm file is added to generate the static library, which size is 320kb. In both static library project and program project,FPU32、TMU is open , floating point mode is relaxed. Actual running tests found that only 2.8 ms is required to call the matrix calculation function in the static function library generated by CCS3.3, however, the same matrix calculation function takes 3.3 ms in the static library generated by CCS6.1.
Why does the same code increase both the static library size and the function call time in the higher versions of CCS and Code Generation? How to reduce running times of matrix calculation function in higher versions of CCS and Code Generation? For me, reducing running times of matrix calculation function is very important.