C66XCODECS: Issue in ti standard library function abs()

Jeslin Nirmala

Prodigy 170 points

Part Number: C66XCODECS

Hi Sir,

We are doing Unit Test in VectorCast where we are building the source code using below compiler

ccsv8

Compiler: c6000_7.4.2

Also using simulator tisim_c64xple.ccxml for running the test cases

We have below code snippet in our source code.

If we pass -2147483648 (int32 min value) for f_inArg0, then abs function is returning as 2147483647 instead of 2147483648

Please check and let us know the solution.

fix16Q16_t Fix16Q16::divideVal(fix16Q16_t f_inArg0, fix16Q16_t f_inArg1, uint32 f_divider, uint32 f_bit) const

{

uint32 l_remainder = static_cast<uint32>(abs(f_inArg0));

}

over 2 years ago

0 Keith Barkley over 2 years ago

Guru 35580 points

In H&S, LONG_MAX (a 32 bit value) is 2^31-1 or 2147483647, LONG_MIN is -2147483647

0 Keith Barkley over 2 years ago in reply to Keith Barkley

Guru 35580 points

In my TI limits.h, LONG_MIN is generally defined as (-LONG_MAX - 1) which is consistent with what you are seeing.

0 Mukul Bhatnagar over 2 years ago in reply to Keith Barkley

TI__Guru* 78325 points

Hi Jeslin

These codec libraries are legacy software products and provided as is. We do not have support on these.
If you are looking for support for video codecs I recommend reaching out our 3P like Path Partner

What device are you trying to leverage these codecs? c6678 or something else?

Regards

Mukul

Code Composer Studio™︎

Code Composer Studio forum

C66XCODECS: Issue in ti standard library function abs()