Effect On dGPU Market and Conclusion
Considering the above points, an APU such as theorized above could massively shrink the potential market for 80$-130$, 100mm^2-150mm^2 GPU’s, currently held by the Radeon RX 460 and the Geforce GTX 1050. With the rising costs of taping out desgins with each subsequent node, if the market for such cards shrinks due to APU’s, it might no longer make economical sense to design and tapeout such chips. Such a situation might force NVIDIA and AMD to use larger dies like Polaris 10 or GP106 found in RX 480 and GTX 1060 respectively, by cutting them down and releasing them to the 80$-130$ segments. It would be easier to amortize costs between, for example, GTX 1050, 1050 Ti, 1060 3GB, and 1060 6GB, all using the same chip, than to have two chips, each having to pay for itself. Margins will go down at the lower end, but the market might still exist for PC’s that already sit on a non AM4 platform, or already have a good CPU without an iGPU.
Another possibility would be said market segment turning into the sub 80$ market, where GPU’s from previous generations are simply recycled and rebranded, rarely if ever being worth the price to consumers, effectively killing the market segment faster than it would otherwise go to Raven Ridge, as opposed to a situation where the segment is kept up with newer architectures and designs.
While Ryzen 7 CPU’s are the ones creating the biggest hooplah right now, and everyone is anticipating NVIDIA’s announcement of a beastly GTX 1080 Ti, Raven Ridge could be the biggest shakeup of the x86 computing market in nearly a decade, all the while bringing a starved AMD some much needed margins and volume. Raven Ridge truly is the potential silent killer not many thought about.