Computex 2018 brought with it a huge announcement for computer lovers and miners alike – the world’s first ever 7nm GPU. This came in the form of the VEGA INSTINCT, a 32GB beast that has already got tongues wagging. Whilst the VEGA INSTINCT has been released to compete in the machine learning market, AMD’s head Lisa Su has confirmed that consumers can expect the architecture commercially in the future as well. This, in conjunction with AMD promising a new Graphics product every three years bodes well for the GPU market in general.
What is Infinity Fabric?
Infinity Fabric is not new to the AMD stable. It has been designed to facilitate communication between components and has been showcased in AMD processors previously, specifically the Threadripper and the Ryzen. In the Instinct, AMD have extended the role to be played by the Infinity fabric, to facilitate better communication pathways between multiple GPUs. This is not a new development, with something similar seen in top of the range NVIDIA products. Infinity Fabric has also previously been seen facilitating communication between components like GPUs and your machine’s hard drive. Popular consensus is that latency is reduced as a result of the infinity fabric and it is part of the logical progression for this tech that it will soon facilitate communication between CPUs and graphics cards. AMD could be the best positioned to make the move towards this development, which will excited miners everywhere.
Timeline And Possible Release Date
The expectation for AMD in 2018 was a release of its 12nm GPUs. Thus, the acceleration to 7nm processes is a big one, which should excite the community. The Vega is currently being tested with AMD’s partners, with an expected release date of sometime towards the end of this year, including several initiatives designed to ensure optimal performance in the machine learning space. AMD’s newest GPUs, complete with 7nm processes, are expected prior to the end of 2020.
What Specifications Have Been Promised
Whilst actual specifications have not been confirmed, the Radeon Instinct Vega is expected to double the process thickness of its predecessor, whilst coming in at roughly 60% of the size. Importantly from a mining perspective, the Instinct doubles the power efficiency of the 14nm GPU that was previously AMD’s flagship, but improving on performance by 35%. These expected specifications are a clear indication that the 7nm process being developed by AMD is nearing completion, a theory which is backed up by the fact that AMD’s 7nm CPU was also demoed at this year’s Computex. AMD have also promised that their GPU comes standard with deep learning functions.
The Radeon RX Vega Instinct in Action
The power of the latest and greatest of AMD’s GPU stable was shown running a Cinema4D render, making use of AMD’s very own freeware app, Pro Render ray tracing solution. With the ever-increasing memory requirements of GPUs in order to facilitate AI architectures, AMD’s newest showcase required all 32GB of the HBM2 that it was equipped with. Whether or not AMD also plan on releasing a 16GB model is unclear, but the 32GB model coped well with the tasks bestowed upon it.