TMS320F28377D: Using the standard atan fucntion

UserHomeInit

Dear TI hello,

We have an algorithme using muttiple of atan(x) function, this function was simplified because we use and old DSP.

I am uisng a double core 28377D floating point now and, I am not sure if there is an alternative to standered function or TI function uisng FPU.

Is it safe to use it, do I have to activate any flag in CCS?

How about the performance, is it greedy in terme of clock cycles? I have a real time system with motor control so it is very important to have an idea about the time it tooks to calculate any value.

Thank you in advance for help,

S.T

over 3 years ago

0 Sira Rao80 over 3 years ago

TI__Mastermind 28155 points

Please set --fp_mode = relaxed under compiler options, and the compiler will replace atan() with the ATAN TMU instruction (see spruhs1, section 7.5.3). This instruction is much faster (a few CPU clock cycles) than the Compiler RTS library's implementation of this function.

0 UserHomeInit over 3 years ago in reply to Sira Rao80

Expert 2225 points

Hello Sira,

Is the same thing with atan2(x,y), I could see in section 7.5.3 that replaced atan is a PU Atan(x) and not Atan(x)?

0 Sira Rao80 over 3 years ago in reply to UserHomeInit

TI__Mastermind 28155 points

The compiler will perform the necessary instructions for the intrinsics to work as expected. Refer to 7.6.3 of spru514, relevant portion pasted below.

C2000™︎ microcontrollers

C2000 microcontrollers forum

TMS320F28377D: Using the standard atan fucntion