AMD unveiled the much-anticipated Vega GPU at CES 2017. The chip giant presented an overview of the Vega architecture and revealed new key features and technologies. Vega 10 is expected to handle memory traffic much more efficiently. AMD also announced that the new architecture will significantly cut back on wasteful memory allocations and showed a demo with two examples wherein the High Bandwidth Controller reduced wasteful memory allocations by as much as 50%. With several new technologies in place, Vega GPU is all set to give tough competition to Nvidia’s upcoming GeForce GTX 1080 Ti.
Nvidia GeForce GTX 1080 Ti may be slower than AMD Vega
Nvidia GeForce GTX 1080 Ti, the last card to come with the famous and well-established Pascal architecture, boasts of 12 billion transistors and is expected to come with 10GB RAM. The new card from Nvidia will attempt to bridge the gap between the powerful GTX 1080 and the flagship Titan X. GeForce GTX 1080 Ti will not come with GDDR5X that featured both on the GTX 1080 and GTC Titan X. It’ll instead go with GDDR5 featuring 348-bit memory bandwidth. Interestingly, it may still be slower than the AMD’s latest GPU sporting Vega architecture.
The Radeon Instinct MI25 – one among AMD’s new line of machine intelligence accelerators – is based on Vega architecture and provides up to 25 TFLOPs, setting the baseline for Vega at about 45% faster than Fury X. This puts Vega-based GPUs at a performance level that matches Titan X and beats GeForce GTX 1080 Ti.
AMD Vega GPU and Nvidia GeForce GTX 1080 Ti release dates
AMD is expected to launch their high-end Vega GPUs in May. The launch is expected to take place at a special event, likely before Taiwan’s Computex, that’s scheduled for May 30 – June 3, 2017. Nvidia GeForce GTX 1080 Ti, is all set to be released during PAX East 2017, that’ll start in just about a month from now – March 10, 2017. PAX – a forum that draws tens of thousands of gamers every year – could be the right platform for releasing the much-anticipated GeForce GTX 1080 Ti.