// BIT_CONVERT - Theis operator converts between integer and FP values, as
// if one was stored to memory as integer and the other was loaded from the
- // same address. The source and result are required to have the same bit
- // size (e.g. f32 <-> i32). This can also be used for int-to-int or
- // fp-to-fp conversions, but that is a noop, deleted by getNode().
+ // same address (or equivalently for vector format conversions, etc). The
+ // source and result are required to have the same bit size (e.g.
+ // f32 <-> i32). This can also be used for int-to-int or fp-to-fp
+ // conversions, but that is a noop, deleted by getNode().
BIT_CONVERT,
// FNEG, FABS, FSQRT, FSIN, FCOS - Perform unary floating point negation,