Here is the full text as found at The Register:
http://regmedia.co.uk/2015/11/06/amd_bulldozer_lawsuit.pdf
So far a core is expected to execute one instruction at one point in time. 8 cores should do 8 instructions simultaneously according to the document. No word about throughput or latency or if it's still ok, that 8 IDIVs take tens of cycles on each core (but are being executed still).
BTW, beginning with multi cores, there were more and more things to be shared, beginning with the external busses or internal/external memory controllers. L2 caches were shared, L3 later, etc. There is cache coherency traffic, reducing the scaling. And so on.
And in the times of power efficiency, even the power budgets are shared between cores, causing 8 of them not being close to 2 times as fast as 4 cores.
The lawyers will have a lot to learn in this case.