Apr 25, 2024 | Updated: 11:35 AM EDT
May 19, 2017 05:17 PM EDT
AMD would ship in the second quarter of 2017 the Radeon Vega Frontier Edition. With 64 compute units and 4,096 steam processors that can produce 13 TFLOPS of performance, the upcoming graphics card will be the fastest graphics card on the planet, according to AMD.
The Radeon Vega Frontier Edition is higher by 1 in terms of TFLOP of performance compared to the Nvidia Titan Xp. Its memory is 16GB of superfast HBM2 which comes with a memory bandwidth of 460GB/s, GSM Arena reported. Nvidia’s other graphics card, the GTX 1080 Ti, offers 11.3 TFLOPS.
According to Digital Trends, the Radeon Vega Frontier Edition is the first graphics card based on the Vega architecture. It was designed for artificial intelligence, creatives, and science pioneers, and not for PC gaming. AMD calls the card’s High Bandwidth Memory 2 as a cache and not memory or video RAM.
AMD points out that the revolutionary memory engine of the Radeon Vega Frontier Edition can access terabytes of memory. The memory is not part of the video card or GPU, but it can be attached if the system builders want to. In the supercomputing and enterprise arenas, the ability to access terabytes of memory, with some possible customization is one of the key advantages of the Vega architecture.
The company calls the new compute engine of the Radeon Vega Frontier Edition as NCU which enables the use of a double packed math for double throughput using 16-bit data. It has the potential to double the throughput to 26 TFLOPS, although it is a feature intended for very specific uses and not on games or daily applications.
However, Tech Power Up reports that gamers would likely be disappointed with the Radeon Vega Frontier Edition because it would remind players of the dual Fiji. The promising graphics card design was unsuitable for gaming and marketed for creators of content.
