Some time ago we talked about the new AMD APU “Fenghuang”, a response to the Intel Kaby-Lake G processors that also include AMD Vega graphics. In the Intel versions, the new processors based on Kaby Lake architecture include Radeon RX Vega M GL or Radeon RX Vega M GH graphics, which have 20CU and 1280SP, and 24CU and 1536SP respectively, which are also accompanied by 4GB of dedicated HBM2 memory.
For Intel, this is a significant qualitative leap forward from its integrated graphics processors, as the specifications of these iGPUs put the graphics at a higher level than the dedicated Radeon RX 560 and not too far from what the 4GB Radeon RX 570 can offer, although there are significant differences.
AMD opts for a second generation Ryzen quad-core processor for its new APU, which includes SMT to offer eight processing threads, while the GPU has been upgraded to 28CU and 1792 SP the concept included in Intel, so the Fenghuang GPU is necessarily faster by simple specifications. Interestingly, AMD has chosen to accompany its 2GB HBM2-only APU as dedicated memory, which arrives at 2.4Gbps and increases bandwidth to 307.2GB/s, compared to 204.8GB/s for Intel’s best choice. This means that the AMD APU has half the memory, but 50% more bandwidth.
Recently, 3D Mark results from this new APU have appeared, comparing against an Intel Kaby Lake-G with Radeon RX Vega M GH graphics. From the results, we can see how the Fenghuang APU wins notably in the graphical tests, although it loses in the physics tests because its CPU achieves lower scores.
Fenghuang’s APU memory can be a limiting factor for some modern games, although AMD may well end up using a hybrid system of shared memory with the system for its GPU. So passing the 2GB we have the ability to pull the memory slower DDR4 system for games that require a little more.
Be that as it may, the concept is very interesting and I’m sure we’ll see this APU bring many laptops and desktops to life with low gaming expectations if the price goes along with it.