That's how it works () - you still need to make sure in your interpreter that your traces are optimizable by the SIMD optimizer, which is usually relatively f

That's how it works () - you still need to make sure in your interpreter that your traces are optimizable by the SIMD optimizer, which is usually relatively fragile, that's just a feature of the optimizations at hand, GCC/clang also have this problem