Speculations are rife that AMD’s Vega GPU could be released in the first quarter of 2017, much ahead to its earlier time frame of mid-2017.
The forthcoming Vega 10 Chip is built on 24 TFLOPS of 16-bit single-precision performance, accompanied by 16GB of HBM2 and 512GB/s of memory bandwidth with a total count of 64 compute units, reports VideoCardz. The 16GB of HBM2 is still a standard inclusion, but the RAM bandwidth is an aspect that doesn’t stand true to the expectations from AMD. HBM2 has been designed with an intent to support up to 256GB/s of bandwidth per stack, and the first-generation HBM2 solutions will be stacked in 4GB.
AMD’s Vega 10 products, slated to get released in a few months from now, will be an eye-to-eye competitor to Nvidia’s GeForce GTX 1070 and GTX 1080 graphic cards.
The Vega 20 chip, the other variant, features 4,096 shader processors, a factor that suggests the availability of 32GB of HBM2 memory. The chip has a lower power requirement that ranges up to 150 Watts. The 20 variant of the Vega class is designed to support PCI Express 4.0 as well.
In the last, AMD will roll out Navi’s 10 and 11 variants soon, engineered to deliver a highly scalable performance and support the next-gen video memory of HBM3. However, their launch has currently been put on ice by the company.