Lines Matching defs:Widen

4001 /// Widen a vector to a larger size with the same scalar type, with the new
4015 /// Widen a vector to a larger size with the same scalar type, with the new
4029 /// Widen a mask vector type to a minimum of v8i1/v16i1 to allow use of KSHIFT
4039 /// Widen a mask vector to a minimum of v8i1/v16i1 to allow use of KSHIFT and
4291   bool Widen = !(Subtarget.hasVLX() || VT.is512BitVector());
4294   if (Widen)
4312     if (Widen)
4319   if (Widen)
4439   // Widen the vector if needed.
8500       // Widen smaller SrcVec to match VT.
20252   // Widen to 512-bits if VLX is not supported.
20891         // Widen to 512-bits.
20991     // Widen vXi32 fp_to_uint with avx512f to 512-bit source.
21023     // Widen vXi64 fp_to_uint/fp_to_sint with avx512dq to 512-bit source.
24372   // Widen to 512-bits if VLX is not supported.
24729     // Widen the vector, cast to a v2x64 type, extract the single 64-bit element
31180     // Widen the vector in input in the case of MVT::v2i32.
31906 /// Widen a vector input to a vector of NVT.  The
32635     // Widen the result with by padding with undef.
32671   // We might have generated v2f32 FMIN/FMAX operations. Widen them to v4f32.
40076   // Widen any subvector shuffle inputs we've collected.
48562       // Widen input to v16i32 so we can truncate that.
51124       // Widen the result if its narrower than 128 bits.
51637   // Widen v2i1/v4i1 stores to v8i1.
57542   // Widen to at least 8 input elements.
57682   // Widen to at least 4 input elements.
58619         // Widen to 64 bits here to get it sign extended.