ethomaz on 22 September 2013
Good, good.... they testes 14CUs with the same ROPs/TMUs and of course get a worst increase in performance than upping 6% the clock.
Now try to use the same 14CUs with more ROPs/TMUs units... you know there are a bottleneck here... increasing the clock increase the ROPs/TMUs performance too... so clock is a better choice here.
eSRAM yet makes no sense but it is not a big deal.
And the redundants CUs shows how bad they are in production yields.
PS. It's good to know the DevKits are running with 14CUs... that's can explain why the games at E3 were better than the Gamescom and after.
