AMD'S NEXT-GEN GPU architecture dubbed Polaris 30, is set to make its debut at the end of 2018, according to WJM47196, a user on Chinese site ChipHell.
The post notes that Polaris 30 will be manufactured using the 12-nanometre finFET process and will offer a 20 per cent boost in performance over the Polaris 20-based high-end GPUs that debuted in 2017.
But WJM47196 wasn't done with setting the rumour mill spinning on Polaris, as they also noted that Navi architecture-based GPUs, aimed at mid-range graphics cards, will make their debut before higher-end pixel-pushing powerhouses set to use high-bandwidth memory 2 (HBM2).
Unlike Nvidia, which tends to have one graphics architecture that spreads across its entire range of GPU products, AMD's lineup is a bit trickier to grasp.
Vega was thought to be the follow-up to Polaris, but aside from very high-end and expensive HBM2 sporting graphics cards and its integration into AMD's Ryzen accelerated processing units (APUs), there's been no sign of mainstream and mobile discreet Vega graphics cards.
As such, one can expect Navi to fill that gap. But as WCCFTech noted on WJM47196's post, Navi looks to be tackling mid-range cards next, so AMD will need a stop-gap to fill the need for newer high-end GPUs, which is where Polaris 30 comes in.
But the problem with Polaris 30 is that it doesn't look like it'll offer HBM2, meaning it'll likely be an optimisation over Polaris 20, which isn't ideal for AMD given Nvidia's Turing architecture-based GPUs could make their debut next month.
The real performance boost for AMD's desktop graphics cars will likely come with HBM2 sporting Navi GPUs or a 7nm Vega 20 GPU with HBM2 aimed at gamers, if WJM47196 is correct.
All this is some pretty healthy rumour and speculation for AMD GPU fans to chew over, for the time being, as we wait to for more official announcements and reveals from AMD. µ
Now you can watch documentaries about horribly disfigured people whenever you like
Brad to the bone
Being in a minority of one doesn't make you right
WeWork needs a rework