Output Modification

The output data can be modified from its nominal form into a modified form. The nature of the modification depends on the nominal output (a bit-vector, or an array); i.e. whether the bit-width of the output is 1 or more than 1.

The output modification is specified through an enumeration in the field qpl_job.out_bit_width. It can have logical values: 0, 8, 16, or 32. A value of 0 corresponds to no output modification. The other three values invoke output modification.

Output Modification for Nominal Bit Vector Output

In case of nominal bit vector output, when the bit vector is expected to be sparse, it may be more efficient to represent the bit vector as a list of indices of those bits that are 1 rather than as a dense bit vector. In the worst case, if the bit vector is expected to have only a single 1-bit, it would be better to get the index of that 1-bit instead of the large array that is almost all 0.

In this case, if output modification is invoked, then the output is a list of the indices of the 1-bits, where the indices are written as an 8-, 16-, or 32-bit value. If an actual index to be written is larger (in bit-width) than the specified size, the operation terminates with an error.

Output Modification for Nominal Array Output

Here, the nominal output format is a packed array of integers. The output modification is to extend the bit-width of the output to the specified size, effectively “unpacking” them. This may make it easier for software to process the results.

For example, if the output is nominally 11-bits wide, it could be written as 16-bit integers. The specified size must be at least as wide as the nominal sizes.

Warning

User must not truncate an 11-bit wide value to 8-bits.