SFF.Network [SFF Network] AMD's AM4 - A Unified Socket

At CES 2016, AMD has talked a little bit more about its upcoming Zen architecture processors. Specifically, the company revealed the socket that will be used and announced the code names of the new CPUs and APUs. We posted a leak about these names and some specs here.

The current lineup of AMD CPUs and APUs spans across 3 different sockets: The FX line of CPUs drop into the AM3+ socket, the APU lineup uses socket FM2+, and the Athlon and Sempron SoC chips slot into socket AM1. These three sockets exist because each of the processor lines has very different underlying architecture.

Read more here.


Lord of the Boards
SFFn Staff
May 9, 2015
A good strategy as they simplify their lineup and the OEMs will have less products to develop. This will help them regain market value if Zen succeeds in being performance-competitive with Intel.

We're seeing a lot of product optimization with AMD and I like it.


King of Cable Management
Sep 26, 2015
Some more information on the new reference cooler called the Wraith that is less than 1/10th as loud as the out-going cooler:


Can anybody identify the mITX board above?

They have some additional Zen rumor reporting. It looks like they were confusing technologies...HSA is heterogeneous System Architecture, which is where there is coherence across various processors with different instruction set architectures. HBM (high bandwidth memory) is the stack graphics memory. It would be very interesting if AMD had some high performance offerings with HBM on-board. It would not be unprecedented to use graphics memory as main system memory, as that is what is done in the Playstation 4. But again, there appears to be some confusion...


John Morrison. Founder and Team Leader of SFF.N
Original poster
SFF Network
SFF Workshop
SFFn Staff
Jun 19, 2015
The board is http://www.gigabyte.co.nz/products/product-page.aspx?pid=4745#ov (note the 'Designed in Taipei' moniker under the pcie slot)

Re: HBM - Graphics memory isn't ideal for CPU loads - it has a significantly higher latency. This is fine for GPUs as as all the data is preloaded into VRAM, but CPU workloads change constantly and thus ram content needs to change more often
  • Like
Reactions: NinoPecorino