On AVX/AVX2 the type v8i1 is legalized to v8i16, which is an XMM sized
[oota-llvm.git] / test / CodeGen / X86 / neg-shl-add.ll
2010-01-19 Dan GohmanFold (add x, shl(0 - y, n)) -> sub(x, shl(y, n)), to...