[Updated] AMD’s Next-Generation Polaris GPU Architecture Codename Leaked – GCN’s Successor, Powering Pixels and Efficient Graphics Cores
AMD’s recently formed, Radeon Technologies Group, led by Raja Koduri is on a roll with announcements. Their recent list of announcements include the brand new Crimson Drivers for Radeon graphics cards, the GPUOpen Initiative that focuses to open-source tools for games and compute and AMD also recently announced the new GPU roadmap that talks about the new display technologies to be integrated inside the company’s latest GPUs that we will be looking in 2016.
Image Credits/Source: HWBattle
[UPDATED 2015 December 31st 2:30 AM ET] HWBattle have issued an update, in which they’ve taken down the image you see above in response to a request from the leaker of said image according to the site. In its splace, they’ve published a slide from AMD’s GPUOpen announcement presentation referencing the code name “Polaris”.
AMD’s Next Generation, GCN’s Successor Is The Polaris GPU Architecture
While it is known since July that AMD will have a new graphics update coming in 2016, an update on AMD’s GPU technologies has silently came from HWBattle. According to the site, we can now confirm that AMD is finally readying a massive update to their GCN GPU architecture which is known as Polaris. The Polaris GPU architecture is not the name of a single graphics chip but an architecture that will be featured across a line of graphics cards that will be arriving in 2016. You might have already heard about some possible codenames of these graphics cards along with some nifty details but the package as a whole will be built on the foundation of the Polaris architecture which is keeping two things in mind, better Pixels and greater efficiency. The banner image has the following to say:
Introducing the Polaris Architecture
Our guiding lights is to power every pixel on every device efficiently. Stars are the most efficient photon generators of our universe. Their efficiency is the inspiration for every pixel we generate.
AMD is hinting a lot in this small statement. First of all, what is Polaris? Polaris is regarded as the brightest star in the constellation Ursa Minor and it is the 50th brightest star that is viewable from our Planet Earth. But what does that have to do with GPUs? A few months ago, AMD announced that they will be bringing better Pixels to the PC platform. Like stated, AMD wants to think of Stars as their inspiration for bringing Pixels efficiently on a range of devices. This is where Polaris comes in, the new architecture that will not only deliver better efficiency than all previous generations combined, but also bring better viewing experiences, powering greater Pixel view on displays.
But that’s not all, talking about efficiency, AMD is bringing 2x performance per watt on their next FinFET based GPUs that will be based on the Polaris GPU architecture. These new high-performance, enthusiast focused GPUs will use the 2nd generation of High-Bandwidth Memory which will be tuned to operate with 1 TB/s bandwidth speeds. We have already learnt that AMD plans to bring HDMI 2.0a, DisplayPort 1.3 functionality on these GPUs and better Free Sync support with FreeSync over HDMI that will be available start in Q1 2016.
To provide consumers a better experience, AMD is going to fully support AR/VR (Augmented Reality and Virtual Reality) on their entire line of graphics cards. Better Pixels is just the tip of the iceberg as AMD has also officially stated that they plan to provide better HDR to everyone along with larger color space to consumers. The basics are that AMD wants to bring 10-bit color rendering to every type of workflow, especially games and that they also want to introduce the world of true HDR to consumers. You can find a detailed coverage of AMD’s talks regarding bringing better Pixels to PC and their 2016 GPU roadmap in their respective articles.
AMD is definitely trying hard to up their game in the graphics department with the newly formed RTG (Radeon Technologies Group). AMD has lost a good share of the discrete graphics market to competitor NVIDIA and will try to gain it back with the new GPUs powered by the latest Polaris Architecture. The announcement on these new GPUs could be expected shortly around CES 2016. NVIDIA themselves will be at the stage of CES 2016 for their own conference where they are expected to announce some details regarding their own next generation Pascal GPU architecture which will be powered with the FinFET architecture. Expect a heated battle between both companies in the summer of 2016 when new, enthusiast-grade graphics cards are expected to launch.
AMD and NVIDIA Next Generation GPU Architecture (Speculation):
|GPU Family||AMD Vega||AMD Navi||NVIDIA Pascal||NVIDIA Volta|
|Flagship GPU||Vega 10||Navi 10?||NVIDIA GP100||NVIDIA GV100|
|GPU Process||14nm FinFET||7nm FinFET?||TSMC 16nm FinFET||TSMC 12nm FinFET|
|GPU Transistors||15-18 Billion||TBC||15.3 Billion||21.1 Billion|
|GPU Cores (Max)||4096 SPs||TBC||3840 CUDA Cores||5376 CUDA Cores|
|Peak FP32 Compute||12.5 TFLOPs||TBC||12.0 TFLOPs||15.0 TFLOPs|
|Peak FP16 Compute||25.0 TFLOPs||TBC||24.0 TFLOPs||120 Tensor TFLOPs|
|VRAM||16 GB HBM2||TBC||16 GB HBM2||16 GB HBM2|
|Memory (Consumer Cards)||HBM2||HBM3||GDDR5X||GDDR6|
|Memory (Dual-Chip Professional/ HPC)||HBM2||HBM3||HBM2||HBM2|
|HBM2 Bandwidth||480 GB/s (Instinct MI25)||>1 TB/s?||732 GB/s (Peak)||900 GB/s|
|Graphics Architecture||Next Compute Unit (Vega)||Next Compute Unit (Navi)||5th Gen Pascal CUDA||6th Gen Volta CUDA|
|Successor of (GPU)||Radeon RX 500 Series?||Radeon RX 600 Series?||GM200 (Maxwell)||GP100 (Pascal)|