Lines Matching full:vpermilps
8925 // If we have AVX, we can use VPERMILPS which will allow folding a load
9229 // If we have AVX, we can use VPERMILPS which will allow folding a load
11373 // two 128-bit lanes use the variable mask to VPERMILPS.
24900 // AVX introduced the VPERMILPD/VPERMILPS float permutes, before then we
24946 // We need a repeating shuffle mask for VPERMILPS/PSHUFD.
25233 // the 128-bit lanes use the variable mask to VPERMILPS.
