Lines Matching full:transformation
3592 // logical operation, do not perform the transformation. The highBitSet
4308 // transformation. Returns true if extension are possible and the above
4309 // mentioned transformation is profitable.
4364 // a good reason for the transformation.
4459 // on vectors in one instruction. We only perform this transformation on
4752 // on vectors in one instruction. We only perform this transformation on
4900 // transformation.
4982 // on vectors in one instruction. We only perform this transformation on
5130 // This transformation isn't valid for vector loads.
5215 // transformation is not equivalent, and the downstream logic to replace
5642 // do that. Clean up the old node and forego the transformation.
5940 // We don't need test this condition for transformation like following, as
6884 // Restore N1 if the above transformation doesn't match.
7019 /// transformation, the new indexed load / store has effectively folded
7155 // not profitable to do this transformation.
7250 /// The transformation folded the add / subtract into the new indexed
7319 // transformation.
7762 /// deems the transformation profitable.
8814 // transformation in other cases anyway.
9682 // Do not let this transformation reduce the number of volatile loads.
9694 // src value info, don't do the transformation if the memory
