5 * Important: iropt: Make sure XorV128 and XorV256 of identical
6 args gets folded to zero
8 * add more iteration in test cases
10 * math_UNPCKxPS_128: use xIsH ? InterleaveHI32x4 : InterleaveLO32x
11 I think this is safe w.r.t. the backend
13 * math_UNPCKxPD_128: ditto
15 * math_UNPCKxPD_256: split into 128 bit chunks and use math_UNPCKxPD_128
21 * for many (all?) of the vector shift-by-imm cases (pre-existing as
22 well as AVX), out of range shifts are not handled properly and only
23 work I think because the host happens to have the same semantics.