I am using Blackfin built-in arithmetic intrinsic function, add_fr1x32() in my algorithm and found that the compiler is generating non saturation instruction (NS). CCES version is 2.5.1 and BF607 processor. I've also observed the same behavior with VisualDSP++ as well on BF537 processor. This is very urgent for my customer and could anyone please reply as early as possible? Is there any compiler switch to select saturate or non saturate version?