The DDS Compiler sine and cosine
outputs are generated as unsigned integer or single precision float values that are cast
to doubles on the interface. To restore the sign for printing to stdout or use outside the C model, you must either cast back to integer
and scale accordingly for the integer output type, or cast back to float for the float
output type. Because both sine and cosine can be a maximum of 26 bits wide (integer),
casting to an int is sufficient. To scale the data, it is recommended that the data is
shifted up by the difference between the width of an int (usually 32 bits) and the
config.Output_Width C model parameter to set the
sign bit correctly, then scaled down by the same value, maintaining the sign.
For example, for a 12-bit output, the scaling value is (32 - 12) = 20, and the correctly-signed sine output can be printed using the following code:
std::cout << "sine " << ((int)sine_output << 20) >> 20) << std::endl;See the C model smoke test, run_bitacc_cmodel.c, for a generalized example.
The phase output from the C model can be up to 48 bits wide, and can be either
signed or unsigned, depending on the configuration. See Output PHASE Channel for more details. In the general case, for a signed phase output, cast the double to
a 64-bit integer type (for example, int64_t), and for an unsigned phase
output, cast the double to an unsigned 64-bit integer type (for example,
uint64_t).