We are potentially days away from a seismic shift in graphics hardware. Both AMD and Nvidia have been teasing their next-generation GPUs for months now, but its Team Red which looks as if it’s making the first move at GDC on March 14th. A live Radeon event is set to take place, during which RTG president Raja Koduri is expected to showcase Polaris to the masses.
It will represent the first graphics card die shrink for almost four years. Due to manufacturing problems the 20nm process was skipped entirely by AMD and Nvidia, so what we’re effectively about to witness is a leap two generations ahead from our current hardware. The expected 14nm process means a die shrink, leading to improved power consumption, more transistors, and a lower price to performance ration.
First of out the gate is expected to be AMD with its Polaris architecture. We’re expecting a mammoth jump in performance here, aided by the strides AMD has made in optimising for DirectX 12. For the first time in a long time it feels as if AMD has beaten Nvidia to the punch.
Not to be outdone though, Nvidia is attempting a similar with Pascal. While details are scarce, this is going to by a die shrink down to either 14nm or 16nm, with an expected triple the memory bandwidth of current-gen Maxwell cards, double the performance per watt and twice the peak single precision compute performance.
Both are anticipated to make use of second generation HBM2 for the on-board memory. You’ll remember first-gen HBM was used with AMD’s Fury graphics cards, but this second generation offers twice the memory capacity and twice the bandwidth. Expect to see terabyte plus memory bandwidth in this upcoming generation.
Both contenders are poised and the scales seem evenly balanced, but who will be declared the winner of the upcoming generation when it’s all said and done? Get voting and let us know your thoughts below!