Radeon Pro Duo
Constructed as a multi-purpose graphics card with an exceptional amount of high computing performance, the AMD Radeon Pro Duo is a GPU that’s designed for the gamers who create and creators who game. Built on the same 28nm Fiji architecture as AMD’s flagship graphics card – the R9 Fury X, the Radeon Pro Duo features two GPU chips on a single PCB, effectively two R9 Fury X’s in a single package. This functions in the same manner as two R9 Fury X graphics cards in a multi-GPU configuration, delivering twice the performance as a single graphics card.
Featuring 8192 graphical processing cores with a GPU clock speed up to 1000MHz,and 8GB of HBM memory running at 1000MHz on a dual 4096-bit memory bus, the Radeon Pro Duo has a total transistor count of 17.8 billion, allowing for 16.3 TFLOPS of computing performance. Requiring 350W of power the graphics card is as power efficient as it high performing, giving the gamers who demand nothing but the best for a high-end gaming experience the ultimate solution.
Designed for 4K gaming, maximum visual quality and high framerates in all the latest titles, the Radeon Pro Duo takes full advantage of the latest DirectX 12 and Vulkan APIs, providing an additional performance boost for games which utilize it. For titles such as Battlefield 1, Sniper Elite 4, DOOM and Rise of the Tomb Raider, gamers can expect faster framerates and consistent performance compared to running the same game under DirectX 11. Working in conjunction with FreeSync capable displays which eliminate stuttering and screen-tearing by synchronizing the display’s refresh rate and the GPU’s framerate output, gamers are ensured a consistent and high-performance gameplay experience.
For Gamers And Developers
Featuring the same cooling solution as the R9 Fury X, the Radeon Pro Duo has an all-in-one watercooler fitted to both GPU chips, in order to provide effective dissipation of heat and stable performance. This also allows the graphics card to run quiet while still maintaining high clock speeds, as needed for high resolution gaming as well as VR experiences. As the graphics cards has been built for developers as much as it has for gamers, the Radeon Pro Duo is also ideal for workstation computing applications where the AMD FirePro Series graphics cards would commonly be used.
Allowing for faster performance, more productivity and improved optimization in workstation applications, the Radeon Pro Duo serves as the ideal solution for developer platforms, where supported applications in OpenCL such as the Adobe Suite and Blackmagic Studio take full advantage of Radeon Pro Duo’s HBM technology and high-performance graphical processing cores. Featuring architectural improvements as well software support for the AMD GCN architecture, applications specifically developed for VR will see increased performance and an improvement to workflows, allowing for faster developments in less time.
Taking advantage of the latest video memory architecture as the R9 Fury X, the Radeon Pro Duo has been built with High Bandwidth Memory – also known as HBM. Where most graphics cards utilize GDDR5 RAM, the Radeon Pro Duo’s HBM provides substantial benefits. Occupying less physical space on the GPU die allows for a greater amount of RAM chips to be fitted on the GPU, and can do so without increasing power consumptions. Where a standard GPU has memory chips built around the GPU die, across the entire PCB – HBM memory is stacked on top one another, sitting directly next to the GPU, on the same package.
The benefits of this design resides in the way that data is transmitted through the GPU die and the HBM itself, which is far more efficient due to the close-proximity of each component making up the chip. As standard graphics cards are made up of a vast number of memory chips, each transmitting data through the PCB by the way of connections and traces, the only way to increase the amount of available memory is use denser or additional chips – each taking up more physical space with an increase in power consumption. This also means that in order to have the memory perform at faster speeds, power consumption rises again, as does heat.
HBM differs from this and not just within the physical design of rearranging the memory chips. As the memory chips are in closer-proximity to the GPU die, it transmits data at a vastly slower speed compared to GDDR5 memory. At first glance, this seems like a downfall, but as each HBM memory chip has a Bus Width of 1024-bit – compared to the 32-bit chips on GDDR5, it is able to transmit a larger quantity of data, making up for its speed – providing 60% more memory bandwidth over GDDR5 . This new memory technology also gives way for physically smaller graphics cards, proving ideal for small-form-factor cases.
Radeon Pro Duo Average Benchmark Results:
Battlefield 1 – Ultra Quality – 4K: 60-70fps
Rise of the Tomb Raider – High Quality – 4K: 60fps
DOOM – Ultra Quality – 4K: 60fps
Grand Theft Auto V – Very High Quality – 4k: 60fps