Looking at a few slides and reading comments on the Ars article, I think the error here is Ars' use of "i.e.". The part probably is 8 (micro-op) issue at execu

Looking at a few slides and reading comments on the Ars article, I think the error here is Ars' use of "i.e.". The part probably is 8 (micro-op) issue at execute, with 3 (fused macro-op) wide decode and 5 wide retire. So the "8-issue" claim could be accurate, but the way Ars presents it is just totally clueless.