Compiler/TMS320C6455: C64+ sign extended right shift intrinsic

Hello,

I am working for Tms320c6455 and hit a blocking issue when trying to do a simple 16bit complex multiplication.

(1) how to force sign extended right shift?

code:

~~~~

long long tmp_pm0, tmp_pm1, tmp_pm;

        tmp_pm0 = _cmpy(_loll(_amem8_const(&a[i])),  _loll(_amem8_const(&b[i])));
            _pack2(_loll(tmp_pm0) >> 15, _hill(tmp_pm0)) >> 15);

I tried casting the each of the _lo & _hi elements to signed before the shift right, still it would only treat is as unsigned and use the SHRU instruction and not the signed SHR.

(2) I thought I could use the intrinsic _shr() directly instead of using the C operator >>, but then _shr() does not exists, and it appears a good portion of the assembly instructions do not have intrinsic equivalents. I thought it would be useful to have the full set.

Please send me suggestion to get sign extended right shift in the most efficient way for 16-bit complex multiply operation.

Thanks,