Lines Matching full:blending

8725 // TODO: Utilize pshufb and zero mask blending to support more efficient
10641 /// This doesn't do any checks for the availability of instructions for blending
12116     // FIXME: Maybe handle this here in case during blending we end up with one?
14326   // a PSHUFB in the end. But once we start blending from multiple inputs,
14353       // We can use an unpack to do the blending rather than an or in some
14443   // Handle multi-input cases by blending/unpacking single-input shuffles.
14622     // First try just blending the halves of V1 or V2.
14666 /// back together vs. extracting the single-input shuffles and blending those
14700   // split them rather than blending because the split will decompose to
16598       // for masking/blending then decompose into 128-bit vectors.
17229     // Try using bit ops for masking and blending before falling back to
29526   // constant plus a BLENDing shuffle instead of scalarizing it.
29534   // lowered as X86ISD::VSRLI nodes in parallel before blending.
45699   // Dynamic blending was only available from SSE4.1 onward.