Speculations suggest that Advanced Micro Devices (AMD) may release its much-awaited Vega 10 graphics processing units (GPUs) during the early parts of next year, to be followed by the introduction of the next-generation cards, Vega 11 during the second half of 2017.
According to WCCFTech, the rumored Vega 10 is seemingly based on the GCN graphics architecture by AMD. It is believed to be composed of 4096 V9 cores that could provide up to 12 teraflops of computing power and the highly sought-after 4K capability. It may also be built using the 14 nm FinFET manufacturing process, with 16 GB of second-generation high bandwidth memory (HBM2) per GPU, a 2048-bit memory bus and a bandwidth of 512 GB/s.
There is also hearsay that there could be a dual-GPU graphics card based on the Vega 10 architecture. The processor is said to be aimed at the professional market initially, for consumers who utilize a high level of content creations. It is said to feature a powerful 128 compute units and 8192 steam processors.
Moreover, it may be packed with 16 GB of HBM2 and marketed with a total memory bandwidth of 1 TB/s. It could also wield 24 teraflops of 16-bit half-precision computing power or floating point performance.
On the other hand, Vega 11 is deemed to be less powerful than Vega 10. An alleged RX 580 GPU is apparently based on the said architecture and possibly offers 8 GB of HBM2, 7 teraflops of computing power, a 1025-bit memory bus and a 256 GB/s bandwidth.
Both Vega 10 and Vega 11 are expected to compete with NVIDIA flagship Pascal-based video cards, the GTX 1080 and GTX 1070. Although AMD usually caters to mainstream and low-end markets with their Polaris lineup, the arrival of the Vega series is deemed to take on the two aforementioned cards.
Since there is no official announcement yet from the developers, available information should be taken with a grain of salt.