In the Investor Day AMD slides now show that AMD shared more details about it's Zen-generation architecture. The biggest factor would be IPC performance being up-to 40% faster per clock cycle.

Zen is the new architecture for the high-end AMD FX line processors, we mentioned this a number of times already. This new FX line will be released in 2016, so they are starting marketing rather soon. The completely new design will be 40% faster per core / instruction / clock cycle compared to the current Excavator cores, and that would be a serious increase alright. The throughput should increase significantly thanks to Simultaneous Multithreading, this is another name for Intel's Hyper-threading.

AMD FX-processors with Zen cores will be seated on the upcoming AM4 platform in the A series, also to be released in 2016. The chips will be fabbed at a FinFET procedure that is 14nm. Recent infortmation have shown 4 and 8-core models to be trelased with a scalable architecture. The 8-core processor would get with 512KB L2 cache per core and 16 MB of L3 cache divded ontow two 8MB clusters for 2x 4 Core modules. So basically it's a scaled up version of the quad-core version (Bristol Ridge). The GCN based IGP will even support HSA (3D stacked graphics memory). Zen resembles the 'Stars' core design that AMD launched with its Phenom series, but more powerful. The module design that you see on the APUs for example are not present here. Each 'cluster'; is its own core, and on an APU each cluster/module has two cores. So each core should be spicy in terms of performance in what seems to be a very parallel approach. 

The new processor series will support both DDR3 and DDR4 over the integrated memory controllers as well as PCI-Express 3.0, With an expected launch time frame of 2016 it will be interesting to see how well these puppies will perform.

In the presentation AMD stated to focus less on low-end PCs, tables and smart-phones. And more at gaming, data-centres and embedded custom chips.

Check out the slides: