Finally, the quality time with loved ones, the luxurious eating and the limitless lie ins are behind us, and we can all get back to doing what we really love: analysing AMD’s new Graphics Core Next 4.0 architecture, codenamed Polaris, for hints of what 2016 has in store for Team Red.
Will you need a graphics card upgrade to play the best PC games of 2016?
The company’s newly formed graphics division, Radeon Technologies Group, have shared the first details on the new Polaris architecture (effectively version 4.0 of Graphics Core Next). Built with the company’s new memory and power controllers using an unbelievable 14nm manufacturing process, it’ll find its way into graphics cards in mid-2016.
That node shrink is a big deal. Firstly because it means the individual nodes within the archiecture are now 14 billionths of a meter big, less than half the width of the ebola virus. Secondly, and perhaps most pertinently, because both AMD and NVIDIA have appeared stuck on the 28nm node for some time. The shrink down has huge power efficiency potential.
A behind-closed-doors presentation given to Anandtech showed an early Polaris protype card running in an identical system to NVIDIA’s GTX 950, both running Star Wars Battlefront at medium settings and 1080p, both pushing out 60 fps. The Polaris card drew just 83W, while the GTX 950 drew 140W.
AMD are calling Polaris a”a historic performance leap in power per watt,” and while that early behind-closed-doors demo isn’t an empirical benchmark, it certainly hints at the potential within the new architecture. It’s exciting news for notebooks, too – the die shrink means smaller form factors are possible than in previous graphics card generations, and coupled with better performance per watt that makes for an encouraging recipe for mobile GPUs.
There’s good news for video editors or anyone keen to see or play content at 4K and 60 fps, too: Polaris will support hardware 4K h.265 encoding and decoding at 60 FPS, DisplayPort 1.3, and HDMI 2.0a.
It’s interesting that AMD chose a GTX 950, one of NVIDIA’s lower end cards, for that early comparison. One shouldn’t read too much into it at this stage, but a side-by-side comparison with a 970, or better still a 980 ti would have set the cat among the capacitors.
As it is, it suggests that perhaps the company’s planning to roll out lower-end, cheaper models first and build up to the expensive, PSU-melting cards a while after launch. After all, that was the strategy they employed most recently with their Fury cards, released in the latter half of 2015 while the less powerful 2x and 3x series cards launched much earlier.